Here’s what’s on our minds this week.

🎙️ The Part of the Customer Journey You No Longer Control

Most companies recognize shifts in customer behavior, but only act on them after their current model starts to break down.

In 2022, as LLMs started collecting data from the web, G2 chose to open their review data for AI training, while their competitors kept theirs locked down. With AI actively changing how people discovered products, G2 found themselves at risk of losing direct traffic, control, and monetization. Two years later, G2 is a top-ten cited source in major models, with traffic up 31% year over year.

The companies that kept their data private are now trying to regain the reach they thought was secure. Eric Gilpin, President of Go-To-Market at G2, calls this idea “future always beats business,” but in reality, it’s a tradeoff. You have to weaken your current model before it breaks, so you can be part of where customer behavior is heading.

This is where most revenue organizations get stuck. Systems like pipeline coverage, expansion goals, and Customer Success (CS) capacity are built on the idea that engagement happens inside the product. But customer behavior is already shifting. People are evaluating AI tools, validating in communities, and showing early interest in places your metrics can’t track.

G2 bet that the most important signals would be outside their control, and that being present there was more important than protecting what they already had. Most Customer Success teams are still focused on tracking what happens inside the product, measuring a part of the journey that’s already over.

Here’s the Full Episode:

Watch the full episode right here or on your favorite podcast app, and listen for these two things:

1. Why early discovery is moving outside traditional channels and what that means for when your team can actually influence the outcome

2. How G2 is adapting to answer engines and what it reveals about where customer decisions are really happening

🎯 The [Un]Churned Take: Control Is Slipping Faster Than Capability Is Improving

Customer Success was built for systems it could observe and influence. That assumption is starting to break. When the head of Amazon Web Services suggests models like Claude could replace large parts of SaaS, the implication goes beyond just product disruption. This means that more of the work customers care about is being executed inside systems CS doesn’t fully see or control.

That constraint is already shaping how Anthropic operates. Holding back model releases in a competitive market reflects a simple reality: capability is advancing faster than it can be reliably integrated. The system can do more things, but you can’t count on it to do them the same way every time. For CS, that means the connection between what the product does and what the customer experiences becomes less direct, and harder to manage.

Operators are feeling this before they can fully explain it. Leaders like Nick Mehta describe the current environment as unusually difficult, but the pressure isn’t coming from a single shift. Distribution, product boundaries, and execution are all moving at once, and CS sits at the intersection of all three. The actions that drive customer outcomes are becoming more powerful, but also less predictable, and that tradeoff shows up first in post-sales.

Large systems have always required coordination before they deliver results, but the timeline has collapsed. Missions like Artemis II depend on years of controlled iteration before launch. AI systems are being deployed into production environments in real time, without that buffer, and with immediate expectations for impact. The complexity is similar, but the tolerance for failure isn’t.

CS now operates in a system where it’s harder to read signals, connect actions to outcomes, and understand where we actually have control. At some point, we need to redesign the function around where outcomes are actually decided, not where they’re measured.

☕ Off Topic: What Else We’re Reading

“Claude, OpenClaw, and the New Reality: AI Agents Are Here—and So Is the Chaos”

As agents take on more of the work, outcomes become less predictable and harder to control because they depend on systems you don’t fully own.

📅 Coming Up

Pulse 2026 — Retention is being rewritten in real time. AI, scale, digital—everything is changing how customers stay. Pulse 2026 is the #1 conference where CS, education, community, and revenue leaders go to figure it out together. Join us May 27-28 in Las Vegas and use code UNCHURNED for a special rate.

Register at gainsightpulse.com →

The New Rules of CS — Gainsight’s ongoing webinar series is tackling the questions CS leaders are wrestling with right now. Three sessions are already on demand, and one more is coming up: Turning Risk Signals Into Action.

Catch up on the full series here →

Wrapping Up

Efficiency inside the system doesn’t matter if the decision happens outside it. So the question is: where is your team actually influencing the decision, and where are you just reacting to it?

See you next Tuesday 🧠

🎧 Listen to more episodes of [Un]Churned

Here’s what’s on our minds this week.

🎙️ Building Post-Sales From Scratch Looks Different Than You Might Expect

Ghazi Masood, CRO of Replit, doesn’t have any CSMs. Instead, he relies on “product advocates” who handle onboarding, retention, and expansion. Each advocate works with a field engineer who manages technical details before and after the sale. In their usage-based model, if a customer stops building, they stop paying. So, the whole post-sales approach focuses on delivering ongoing value, not just hitting renewal dates.

In episode 179 of [Un]Churned, “Why Replit’s CRO is Hiring 200 GTM People in 12 Months ft. Ghazi Masood,” host Josh Schachter talks with Ghazi about building a post-sales approach at a company that grew from $2M to $150M in revenue in less than a year. Ghazi’s model combines roles that most companies keep separate. He values product passion more than sales experience and focuses on delivering new use cases to keep customers, instead of relying on QBRs and health scores.

For CS and revenue leaders, the real question isn’t whether to copy this model. It’s about recognizing which old assumptions from a seat-based, renewal-focused approach no longer make sense.

Here’s the Full Episode:

Watch the full episode right here or on your favorite podcast app, and listen for these two things:

1. How Ghazi approaches account segmentation and book-of-business ratios when the whole user base is non-technical, and how this affects hiring decisions.

2. What Replit is doing to nurture a culture around building internal tools, and the impact it’s made on post-sales discussions.

🎯 The [Un]Churned Take: Your Agents Are Only as Good as Your Context

Last year, Shopify co-founder and CEO Tobi Lutke made waves with his argument that context engineering would replace prompt engineering as the core skill in AI. That shift is happening, but it requires a change in how teams work with their data. Agents are only as useful as the context they can draw from, and in Customer Success, the most valuable data has never lived in a system.

And as Artemis II flies closer and closer to the moon, we can’t help but recall when Cognizant CIO Neal Ramasamy coined the term “dark side of the moon” to describe the relationship context that never makes it out of a CSM’s head and into a health score.

The instinct is to connect more systems and feed agents more data. Research suggests that’s the wrong move. And why’s that? As context fills up, recall accuracy drops, even on models with large windows.

The better instinct is curation: what does an agent actually need to know to take the right action on this account, right now? But curation only works if the right signals exist in the first place, including the unstructured ones. The relationship context that used to live only in a CSM’s head now has a path into the system: emails, Slack threads, and meeting transcripts.

Gainsight CEO Chuck Ganapathi framed the design principle in a recent LinkedIn post: “Agents don’t need dashboards and buttons; they need context and the ability to act.”

Gainsight’s MCP launch last week reflects the same logic. By connecting structured customer data from Gainsight CS with the unstructured relationship signals from Staircase AI, agents draw on a curated, unified picture of every account, not a firehose of disconnected data.

If you’re curious where your context layer currently stands, here’s a useful exercise. Take a high-stakes prompt such as “summarize the renewal risk across my top 20 accounts” or “flag accounts where stakeholder engagement has dropped in the last 30 days.” Ask honestly whether your current data could answer it. If the answer is no, you’ve found your starting point.

See what’s possible with Gainsight MCP: explore use cases and prompts →

☕ Off Topic: What Else We’re Reading

“How You Could Be Using AI in 2026” by David Arnoux (LinkedIn Post)

Don’t skip the comments. There’s a strong thread around how teams are still stuck in “prompt mode,” while the real value comes from wiring AI into actual systems and processes, not just experimenting with it.

“Why AI Is Changing How SaaS Products Are Designed” by Fintech Global

SaaS is quietly shifting from building features and workflows to building systems that interpret intent and take action. For CS and revenue leaders, you’re no longer driving usage, you’re on the hook for proving the system is actually doing something that matters.

📅 Coming Up

Pulse 2026 — Retention is being rewritten in real time. AI, scale, digital—everything is changing how customers stay. Pulse 2026 is the #1 conference where CS, education, community, and revenue leaders go to figure it out together. Join us May 27-28 in Las Vegas and use code UNCHURNED for a special rate.

Register at gainsightpulse.com →

AEO Unplugged — Join us on April 9 for a half-day virtual event where CS, community, and education leaders will share what AEO actually means for post-sales teams. You’ll learn from real experts, see what it takes to move from fragmented content to a connected ecosystem, and walk away thinking like a builder.

Register here →

The New Rules of CS — Gainsight’s ongoing webinar series is tackling the questions CS leaders are wrestling with right now. Two sessions are already on demand, and two more are coming up: Turning Risk Signals Into Action and Stop Tracking Adoption, Start Predicting Retention.

Catch up on the full series here →

Wrapping Up

Ghazi’s team didn’t wait until the system was perfect. They built the model, filled the gaps, and kept moving. It’s a good instinct to follow.

See you next Tuesday 🧠

🎧 Listen to more episodes of [Un]Churned

Here’s what’s on our minds this week.

🎙️Betting That Human Engagement Is the Competitive Edge

Cassie Vaughn is the RVP of Customer Success at monday.com and a serious F1 fan. Ask her about the sport, and she’ll tell you the pit stop is more science than spectacle. It looks like chaos from the outside, but every movement is rehearsed and every role is defined.

It’s a metaphor that maps surprisingly well to how she runs her CS org. This year, her team is deploying agents to handle discovery, optimization, and workflow builds, while committing high-touch CSMs to spending double the time with customers compared to last year. The agents run the repeatable work. The CSMs show up on-site, whiteboard customer workflows end-to-end, and help executives think through what their processes look like in two years when AI removes half the steps. It’s disciplined, deliberate, and entirely by design.

For CS leaders, it’s worth pressure-testing against your own org. When agents take the repeatable work off your team’s plates, where does that time go? You can add more accounts to their book of business, or take a page from Cassie’s book and dive deeper into the accounts that matter most.

Here’s the Full Episode:

Hear the full episode right here or on your favorite podcast app, and listen for these two things:

1. Why Cassie doesn’t trust health scores and what she coaches her team to do instead.

2. The case she made internally to tie CSM compensation to retention outcomes, and what it took to get leadership buy-in.

Join Us at Pulse 2026

The conversation doesn’t stop here. Pulse 2026 is where CS and revenue leaders come together in Las Vegas this May to work through exactly these questions. Register now using the code “UNCHURNED” for a special rate.

Register for Pulse 2026

🎯 The [Un]Churned Take: Information Is Table Stakes. Execution Is the Value.

The outcome-focused CS conversation has been building for a while, and a few recent posts make it easier to see where it’s actually headed.

SaaStr founder Jason Lemkin recently shared that his AI agent sent 150 fully personalized sponsor update emails in a single morning, each with unique task lists, links, and action items for every account. His take was that the agent outperforms most human CSMs at the operational layer. Not because CSMs aren’t good at it, but because that kind of coverage, at that level of specificity, is just not physically possible for a human at scale.

The VC world is arriving at the same place from a different direction. Sequoia, a16z, General Catalyst, and Founders Fund are all converging on a thesis that the next wave of value isn’t in selling software. It’s in selling the work itself.

CS teams are good at finding answers. They’re delivering decks, holding QBRs, and sharing recommendations. But, the real value is in turning customers into operators. If agents handle what’s repeatable, CSMs are free to make judgements, strengthen relationships, and be strategic.

As a result, CSMs are navigating more change at once than at any other point in the function’s history. Their role is being augmented, their metrics are shifting, and they're being asked to build commercial and strategic skills on top of it all. An agent can handle the operational layer, but it can't build trust. If CSMs aren't supported through this transition, customers will feel it. Josh Schachter shared a real example on LinkedIn recently.

His colleague closed one of the largest telecom companies in the world after years of prospecting, because their previous vendor cycled through CSMs every few months and never got them to value. The win happened because CS failed at the other company. That’s the commercial consequence of getting this wrong.

The operational layer is being automated, and the strategic layer is up for grabs. The CS orgs that move first will define the role for the next decade.

☕ Off Topic: What Else We’re Reading

“Wiring the Intelligent Enterprise for the Year Ahead - 2026 Will Be a Year of Reeckoning for AI” by Edward Funnekotter

We’ve had years to test and explore what AI can do, but we’ve reached a point where AI needs to transform from a “cool demo” to true business value. These are four trends to keep in mind as the focus shifts from flashy pilots to scalable systems that deliver measurable outcomes.

Wrapping Up

Outcomes don’t build themselves. The orgs that get this right will be the ones that take the human side of this transition just as seriously as the technology.

See you next Tuesday 🧠

🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

The Best Thing You Can Give Your Customers Doesn’t Come From You

Christina Meng leads the Champion programs at OpenAI. Part of this initiative includes running roundtables for OpenAI’s enterprise customers. She sets the topic, kicks things off, has a few questions ready, and then, for the most part, she gets out of the way.

When she lets champions from different companies talk to each other, something changes. They realize they’re all working through similar challenges, and the conversation goes further than any planned agenda could.

Christina grew OpenAI’s Champion Network from nothing to over 600 members in a year, after nearly ten years doing similar work at Slack. She’s learned that lasting AI adoption depends less on the technology and more on having the right people supporting it at every level.

In Customer Success, the usual instinct is to offer more. More content, more training, more QBR slides. But Christina’s work shows a different approach: maybe the most valuable thing you can give your customers is a space where they can connect with others facing the same challenges.

Here’s the Full Episode

Watch the full episode right here or on your fave podcast app, and listen out for these two things:

1. How Christina measures success now vs. when she started. The shift from attendance metrics to sustained activation is an important distinction for adoption programs.

2. The three champion tiers explained in her own words (executive champion, champion lead, internal champion) and how each one maps to a different kind of influence.

🎯 The [Un]Churned Take: We’re All Living in the Frenzy

The Agent Madness tournament just dropped. 64 AI agent builds, submitted by the community, going head to head for the title of “coolest thing anyone built with AI this year.” It’s chaotic and fascinating and a little overwhelming, and honestly, it’s a perfect snapshot of where we are right now.

Economist Carlota Perez has a name for this moment. In her framework for how transformative technologies spread, there’s a phase called the Frenzy. It’s a gilded age where capital floods in, everyone is building and competing and shouting, and the technology is being brute-forced into existing workflows without the institutional redesign needed to unlock its real value.

The real gains, she argues, come later, in the Deployment Period, when organizations actually restructure around the technology instead of just layering it on top of what already exists. It took 30 years to transition factories from steam to electric power. Not because the technology was slow. Because the humans were.

Here’s the CS connection: this is exactly what Christina Meng described in this week’s episode. The enterprises pulling ahead aren’t the ones with the most impressive agent stack. They’re the ones building internal champion networks, restructuring how their people learn, and investing in the sustained behavior change that makes adoption stick. Turns out, the hard part is rarely the technology.

The frenzy is real, but the Deployment Period belongs to the teams that started building human infrastructure yesterday.

☕ Off Topic: What Else I’m Reading

A few things I’m reading beyond the CS lens.

“Has AI Ended Thought Leadership?” By John Windsor

AI is making it easier to talk about a subject, but John emphasizes that the real value comes from “thought doers.” The scar tissue that comes from mistakes creates granular, personal learnings that AI can’t generate. Are you a thought leader or a thought doer?

“The State of AI Right Now” by Fabian Hediger

Enterprise ambition is running ahead of enterprise execution. Only 25% of organizations have moved more than 40% of their AI pilots into production. That gap is either your biggest threat or your biggest opportunity, depending on what you do next.

Wrapping Up

The frenzy is loud and nobody has a complete playbook yet. I shared an operating model on LinkedIn this week that feels right for this moment: LF-See. Not LFG. Let's. Effing. See. Go all in, stay curious, and see where the experiment takes you.

See you next Monday. 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

The energy wasn’t just on stage. Some of the most interesting moments happened in the conversations between sessions.

While the main focus was Gainsight’s plans for 2026, the conversations around them felt familiar. The same themes I hear in my [Un]Churned conversations with Customer Success leaders kept resurfacing in sessions, hallway chats, and late-night debates.

In many ways, RKO felt like a microcosm of what the Customer Success industry is grappling with right now.

So in this special edition of [Un]Churned, I want to share a few themes that kept coming up throughout the week. To me, they signal some of the bigger shifts unfolding across Customer Success.

Here’s what’s been on my mind.

1. Customer Success has one of the most valuable datasets in the company.

Customer Success leaders already understand how much of the customer story flows through their teams. Adoption signals, renewal conversations, and shifts in sentiment typically surface first in the day-to-day dialogue between CSMs and their customers.

What makes this perspective valuable is how it brings different signals together.

Most teams see only one layer of the customer experience. Product sees usage patterns, Sales sees pipeline and contract value, and Marketing sees engagement.

Customer Success brings all three perspectives together.

CS leaders see how customers use the product, how stakeholders feel about their experience, and whether those signs point to growth or risk. This mix of product behavior, sentiment, and revenue context is rare in other parts of the company.

For years, it’s been hard to put that information to use. It was scattered across conversations, notes, and systems that didn’t connect well. Now that AI can analyze these signals together, the data Customer Success teams have is starting to look much more strategic than before.

2. The gap between people using AI as a toy and people using it as a competitive advantage is growing fast.

It was hard to have a conversation during the week that didn’t eventually touch on AI. Most Gainsters and our customers are already using it in small ways, like drafting emails or summarizing meeting notes. Helpful tasks, but still fairly tactical.

Customer Success sits closer to the strategic center of the customer relationship. QBRs, account handoffs, risk escalation, and aligning stakeholders all require a deeper understanding of what is happening across the account.

Several conversations this week focused on how AI could begin supporting that level of work. Brady Bluhm (Sr. Product Manager, Staircase AI at Gainsight), our resident “agentic power user” led a workshop showcasing how he thinks about customer context in LLMs, how to set up inbox reviews, custom Claude skills, and more.

MCP was an especially hot topic. Once your LLM can access the systems where the Customer Success work actually happens, like timeline entries, success plans, conversation transcripts, and product usage signals, the model begins reasoning with the same context the CSM sees.

That’s when AI becomes genuinely useful in Customer Success workflows.

3. The real question is becoming: what outcomes does this deliver?

Another major subject during the week was value realization. Onboarding and adoption conversations used to focus on features. Now, they’re more about the results those features deliver.

People now look at things like how deeply products are adopted, signs of expansion, feelings about ROI, and early signs that value is growing or fading.

This shift came up often in talks about services and advisory work. As Inna Ekhaus (Sr. Director, Service Products) put it, many companies are now in the “show me the ROI” era, where productivity and revenue outcomes are the main indicators of success.

Taylor Johnston (Sr. Director, Advisory Services at Gainsight) shared a similar view in her session on value frameworks. Customer teams are now expected to show real numbers for the outcomes customers achieve and help them find even more value.

As software becomes easier to build and launch, what sets companies apart is increasingly the results customers actually experience. Customer Success is the key to unlocking that value.

4. Nobody is immune to acceleration.

If there was one macro theme running throughout the week, it was the pace of change.

Chuck Ganapathi (CEO, Gainsight) started his keynote speech with a look back on the ripple effects of AI. I use the term “look back” lightly, because SO much has changed in not just the SaaS industry, but the whole world, since just January. We’re barely three months into 2026 and we’ve advanced to having Open Claw running autonomous workflow and vibe coding so powerful you can have an app in the app store in just a few hours. AI is not only changing the way we work, but it’s changing the course of our futures.

New tools show up almost every week. Product cycles move faster, and customer expectations change along with the technology they’re trying out. Erica Kuhl (EVP & GM, Digital Customer Hub at Gainsight) summed up the mood well, sharing that the fast pace of innovation around Claude and the challenge of building great products while everything is changing so quickly has increased the urgency in the way her team operates.

The rapid development in the AI world is changing how investors think about software, how companies choose vendors, and how teams plan their own roadmaps. You can see this acceleration everywhere. It’s in product development, market expectations, how customers evaluate software, and how teams learn and adapt.

Every function feels it.

5. The teams that stay curious will move the fastest.

While acceleration changes the environment organizations operate in, curiosity determines how teams respond.

The most interesting conversations at RKO came from people who were experimenting. They’re trying new workflows, testing AI tools with real customer data, and asking how their roles might change.

These conversations were rooted in curiosity, and the most curious teams tend to stay close to their customers. They follow conversations wherever they happen—calls, communities, events, and more often now, the social platforms where relationships grow over time.

One of the highlights for me was chatting with Darren McKee about using LinkedIn as open terrain for building multithreaded relationships with both new and existing customers.

In a fast-moving environment, these conversations become a powerful way to learn and improve.

The Bigger Shift Beneath It All

The themes from RKO were not surprising individually. AI. Value delivery. Customer context. Market acceleration. What stood out was how consistently they surfaced across different sessions and conversations.

Each voice brought a slightly different perspective, but together they pointed to the same shift.

Customer Success is becoming central to how companies understand their customers and deliver outcomes. The signals are already there. The question now is what teams do with them.

See you next Monday. 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

How This Former CRO Is Making CS Impossible to Ignore ft. Manish Chawla (This Week’s [Un]Churned 🎙️)

As the industry evolves, the expectations around the role of CS are changing. It’s more important than ever to show how the work connects to revenue outcomes.

On this week’s [Un]Churned episode, Manish Chawla, CCO at PowerSchool, shares how he views Customer Success through a revenue leader’s lens. With years of experience leading revenue teams, he uses a straightforward test: if Customer Success doesn’t drive measurable revenue, it won’t pass executive review.

Manish and I discuss how companies are starting to see Customer Success as a revenue discipline. Now, customer signals, expansion efforts, and accountability are all tied directly to revenue results.

This shift makes it much clearer whether Customer Success is effective.

Start Small, Learn Faster

“AI isn’t a magic wand.” Ankit Aggarwal at MoEngage anchored their copilot project around that idea. Using data they already had, Ankit created an AI assistant for CSMs that removed the need to crowdsource answers from other teammates.

The impact showed up quickly, but it required a shift in mindset. The team had to treat the copilot like a teammate, not just another tool.

Here’s his blueprint for building a copilot that made CSMs faster and helped the entire organization get smarter.

Read the full article →

When Busywork Disappears

Will AI replace CS? Todd Saunders suggests something slightly different.

Much of the operational work around the role is exactly the type of work AI handles well. Meeting prep. Account research. Interpreting customer signals across systems. A large portion of the job sits in that layer.

The comments on Todd’s post were interesting. Some saw relief. Others saw risk. When AI removes the busywork, what actually defines the role?

Read the post →

The Delegation Shift

AI is forcing an operating model decision inside Customer Success. Karthick JL points out what happens when systems begin acting on customer signals rather than just surfacing them.

Outreach can trigger automatically, lifecycle plays can launch, and expansion signals can begin revenue motion. Customer Success teams are moving from copiloting work with AI to delegating parts of the job to it.

Read the post →

Bonus [Un]Churned Episode: Inside SAP’s Scaled Customer Success Strategy

What does Customer Success look like when it has to operate at enterprise scale? In this bonus episode, Carsten Schütz, VP & Global Head of Scaled CS at SAP, shares how his journey from startup CSM to SAP executive reshaped his thinking about scale, specialization, and the real role of AI in customer success.

The conversation explores how large organizations structure digital programs, where human engagement still matters, and how both come together at scale.

☕ Off Topic: What Else I’m Reading

A few things I’m reading beyond the CS lens.

“The Real AI Strategy Is Personal” by Randy Wootton

Randy frames AI adoption through a personal maturity model instead of a technology roadmap. Reading it made me think about where I sit on that curve and where I can be more intentional.

There is Exactly One Way SaaS Can Be Saved by Eoghan McCabe

Eoghan shares a stark look at what it took for Fin by Intercom to get record growth from a place of record lows; all while navigating the AI disruption we’re all feeling. It wasn’t easy and it wasn’t pretty, but it’s possible and Fin is proof.

Skills Are Among the Most Consequential New Technologies by Ethan Mollick
Mollick’s post made me think about Skills a little differently. If AI can help create and test them the way we build software, the pace of learning might start to look very different.

Wrapping Up

AI may remove a lot of the work around Customer Success, but that might make the role clearer than it has ever been.

See you next Monday. 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

While this anxiety is valid (and something I’ve also experienced), it can make you vulnerable to thinking the loudest idea is the highest priority.

Big AI projects like predictive renewals, autonomous playbooks, and automated customer engagement are exciting at first. In brainstorming, they sound like they’ll supercharge your team. But in reality, they’re tough to implement, costly to test, and risky to launch.

Often the most effective starting point for AI in Customer Success isn’t ambitious, but simple.

Start with a copilot. Think of an AI copilot as your AI-powered assistant. It helps your team work smarter by streamlining workflows and supporting the tasks they already do each day. These tools use large language models (LLMs) to create, summarize, and analyze content across data sources.

It’s low-risk, internal, and delivers immediate value to the team. You can start small, evolve gradually, and build on familiar platforms instead of buying new ones. Most importantly, it forces your team to organize its knowledge and learn to work with AI, not against it.

That’s exactly what Ankit Aggarwal, Sr. Manager of Customer Success Operations, did at MoEngage.

The Cost of Not Knowing Where to Look

When I first heard Ankit describe his team’s challenge, it sounded painfully familiar. MoEngage had grown to over 150 people in its CS organization, including CSMs, Professional Services, and Operations.

Regardless of the region, industry, or maturity level, they were all losing time to the same problem: finding answers.

CSMs spent hours digging through Confluence, Salesforce, and Zendesk to piece together customer information or product documentation. The same questions kept surfacing in Slack. Senior CSMs were often pulled into DMs to explain a process or troubleshoot an issue they’d already solved many times before.

The knowledge existed, but the issue was access.

As MoEngage’s CS team grew, so did this problem of finding the right answers. But the solution was forming too. MoEngage had recently created an internal AI Task Force. This cross-functional initiative, backed by the CEO and CTO, explored real AI use cases across the company.

Ankit saw the opportunity immediately and carried out the following steps to pitch the idea to the taskforce:

1. Collected recent CSM meetings using data from Gainsight.

2. Exported the data and used AI to categorize them into categories like support, onboarding, strategic, escalation, etc.

3. Discovered a majority of those conversations fell into the ‘support’ category.

4. Delivered the hypothesis that CSMs are spending a lot of time ‘supporting’ the customers and gained buy-in on the agent.

With all the right stakeholders and resources, they could go beyond just building another dashboard. Ankit could solve the “where do I find that?” problem once and for all.

Building on What Already Existed

When building AI quickly, it can be hard to avoid getting bogged down in details. The good news is, you likely already have the data, processes, and tools you need to get started. Data gathering is often the toughest part of setting up an AI agent, but without good data you risk bad output.

In this case, Ankit found that his work over previous quarters had already cleared the runway for their project. He’d recently centralized the team’s documentation and cleaned up their Confluence space. This became the foundation for his copilot.

He mapped the five most common categories of CSM questions—product, troubleshooting, customer info, process, and industry best practices—and identified where that data actually lived. Then he connected it all:

• Gainsight and Salesforce for non-PII anonymized user data

• Confluence for internal documentation

• Zendesk for help articles and tickets

• Google Drive for QBR archives

• MoEngage’s own product telemetry

Working alongside his IT team, he built the copilot in Google Agent Space. With the groundwork completed, the first version was a cakewalk. It took one day to set up and one week to fine-tune.

No new procurement, no complex architecture, just pragmatic execution with the tools already in place.

Treating AI Like a Teammate, Not a Tool

Ankit started with ten CSMs already experimenting with AI in their workflows. They had the ideal combination of an early-adopter mindset and field expertise.

He gave them a simple challenge: use the copilot as if it were a real teammate. Ask it everything you normally ask each other.

They created a private Slack group to share feedback, posting only the wrong or confusing responses. Ankit also sent out surveys to understand how the CSMs were using the agent and if it was effective.

Each night, Ankit reviewed the shared examples, refined the prompts, reweighted the data sources, and adjusted the response formats.By week two, most answers were accurate. The copilot was already functioning as a single, reliable interface to MoEngage CS’s collective knowledge.

The Results

Within a few weeks, Ankit’s ten-person pilot spread to over 150 users. ​​The workflow was simple. CSMs were already using a Slack channel to crowdsource answers from the rest of their team. Now, anytime someone asks a question, the agent replies to it. This AI copilot became a daily starting point for CSMs and the first place they turned for answers.

Senior CSMs regained valuable focus time, instead of being interrupted by colleagues. New hires were able to ramp up faster thanks to the autonomy granted by the copilot. AI was moving from theory to everyday reality with each submitted question.

These results weren’t just anecdotal. Internal data showed that:

• 77% of users said the copilot was more effective than previous workflows.

• 66% saved 1–4 hours per week.

• Accuracy averaged 4–5 out of 5 across use cases.

The biggest win was that the entire AI culture at MoEngage was evolving. It was no longer something to be curious or hesitant about. It was now something they depended on.

The Blueprint for AI Copilot Success

Ankit’s success wasn’t about picking the right model or tool. It was about executional discipline and starting with a problem that mattered.

His approach offers a blueprint every CS leader can follow:

• Clean your data and documentation first

• Start with a small, internal-facing use case

• Pick early adopters, not skeptics

• Build quick feedback loops

• Measure the time and consistency gains

Ankit also shared this advice: “Most people think that AI is a magic wand which will automate Customer Success. Perhaps, there is a lot of noise on social media promising agents that can replace humans. The reality is that you have to think of AI as an enabler. Start by thinking about how you can integrate AI into your current workflows. The efficiency lies there!”

MoEngage’s copilot made CSMs faster while making the entire organization smarter. It also provided a clear proof point for leadership to further invest in AI.

You Already Have the Ingredients

Every Customer Success organization already has the ingredients for a copilot: documentation, playbooks, customer data, and a team that knows what “good” looks like.

The opportunity is to connect it all and turn that knowledge into a shared advantage. If you’re trying to decide where to start with AI, start there. Teams that learn to combine AI with institutional knowledge won’t just be more efficient. They’ll redefine what it means to scale Customer Success.

About This Article

This article is based on a detailed discussion with Ankit Aggarwal, Senior Manager of Customer Success Operations at MoEngage, on how his team approached their AI copilot. The examples reflect real operating decisions made inside MoEngage’s Customer Success organization. All references to systems, workflows, and data have been sanitized to protect confidential business information.

My dog Jackson had a blast, and it gave me some time to hunker down and catch up on some great content.

B2B Communities Aren’t Dead, Your Outdated Metrics Are (This Week’s [Un]Churned 🎙️)

Jon Wishart (VP, Strategy & Growth at Gainsight) and Brian Oblinger (Strategic Consultant and Host of In Before the Lock Podcast) have had a front row seat to the evolution of B2B SaaS customer communities. From the early days of simple support deflection. To the pandemic-led shift to customer advocacy. And now the “no-click era” of AI.

In this week’s episode of [Un]Churned, we explore what it takes to change the perception of your community from a cost center into a retention investment. We pull on a lot of threads: the metrics that need to retire and the new ones that should take their place, the cross-functional relationships that matter, and the necessary groundwork for a successful community.

If you’re building or maintaining a community, or even just considering it, Jon and Brian will give you an unfiltered look at what it takes to make it a value driver.

Can AMs Replace Customer Success?

Earlier this week, I shared a conversation with a CRO scaling their AI company from under 50 to 200+ GTM hires; with no Customer Success Managers. The comments made it clear: what matters is who owns value.

In usage-based AI models, NRR follows adoption. If revenue is the primary KPI, does value delivery become intentional or incidental?

Curious where you land. Can properly trained Account Managers replace Customer Success, or does mindset still matter more than semantics?

Read the full post →

Think Before You Hit the Gas

Our CEO, Chuck, shared with our team Boris Cherny’s (Head of Claude Code) episode of Lenny’s Podcast in his weekly update, and there’s a reason why this is creating a lot of buzz. Using AI to code is old news. What’s interesting is that we’re already seeing the ripple effects. What happens when execution stops being the bottleneck?

When engineers can ship 20 to 30 requests a day without touching the code, speed stops being impressive and starts revealing things. If your roadmap isn’t clear, you’ll see it faster. If you don’t really understand your customers, you’ll build the wrong things faster.

Watch the full episode →

Moving Fast While Staying Human

In Part Two of the “behind the build” of our Instance Review AI agent, we share what really happened when code became reality. Working with Method Garage, an AI transformation company, we were able to move with groundbreaking speed and go from discovery to prototype in just three weeks.

This article is an honest look at what worked, what didn’t, the feedback we received, and the lessons that stuck. It’s an important reminder that keeping humans involved is part of the design, not a limitation.

Read the full article →

P.S. One of the biggest takeaways from this process is that you can’t afford to skip discovery. Go straight to production and you’ll inevitably miss important context. We talk more about our discovery process (and how it taught us how little we actually knew) in Part One of this series. You can read it here.

Go Wide or Go Up

Nick Mehta shared a sharp take this week on a trap many SaaS companies fall into: getting stuck in the $20K ACV middle ground. Too big for true SMB velocity yet too small to compete up-market. Growth looks strong, but then it stalls.

His framing is simple but uncomfortable. You either go wide and expand your footprint across more use cases, or you go up and build for bigger customers with deeper needs. What doesn’t work is standing still and just hiring more sales reps.

It’s a reminder that ACV growth relies on strong product and positioning decisions.

Read the full post →

☕ Off Topic: What Else I’m Reading

A few things I’m reading beyond the CS lens.

“How to Prepare for the Next Decade” by Scott Barker

This post stopped me in my tracks and brought up memories of Boggle nights with my grandmother. If you’re feeling angst in this “acceleration decade,” this is a helpful suggestion to slow down intentionally.

'“Giving Away Your Legos is Fucking Hard” by Molly Graham

This article is a good reminder that growth always costs you something. The work that built your confidence is usually the first thing you have to hand off if you want the organization to move faster than you.

“Workaholic. You have an unhealthy obsession with AI.” by Ruben Hassid

Is AI actually helping you do more, or is it just “performative productivity?” Ruben shares lessons for AI-holics that bring decision-making and prioritizing back into focus.

Wrapping Up

There’s no shortage of momentum right now. The real work is making sure it compounds in the right direction.

See you next Monday. Until then, stay curious and stay warm 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

IRs are one of the most important activities a CSM can do, but also one of the most time-consuming. With so much room for improvement, it was an ideal starting place for us to apply an AI-powered solution. In just one week, we picked apart the current IR process through a rapid yet thorough discovery process. Here’s a quick recap of our findings:

• IRs involve two completely different activities: technical analysis and value narrative.

• We had an internal expert, Principal CSM Elliot Hulverson, who in his 8 years at Gainsight had found a way to make IRs fast and effective. He was our go-to resource for building an efficient solution.

• We had 80%-90% of the data we needed between an existing config export and usage data from our Gainsight PX integration.

• Our purpose shifted from accelerating the IR process, to enabling CSMs to take on this laborious task that was effectively non-existent.

We brought in Method Garage to run a Blueprint Sprint. One week to map the current state. Two weeks to build a working prototype.

Here’s what happened when we turned code into reality, and the biggest lessons came when the system started failing.

What We Actually Built

After compiling the findings from our discovery process, we started the process of crafting an agentic Instance Review.

The prototype could take a customer’s Gainsight setup and create a clear review in a dashboard, similar to how humans do it today. But could we take a motion that normally takes 8-12 hours of effort and reduce the work without compromising judgment?

Instance Reviews aren’t a new concept and the goal wasn’t to just build a clever internal service. We needed to understand whether the review itself can be meaningfully “agentified” without losing credibility with customers.

We also wanted to create this solution using reusable building blocks, rather than a fixed workflow. Our objective is to eventually use this same engine to support QBR prep, risk response, and health monitoring without having to rebuild logic each time. So we opted to break the work into smaller pieces.

In our discovery process, we had identified the two activities that made up IRs: technical analysis and value narrative. So, we split the review process into two distinct capabilities that aligned with these activities:

Config Intelligence parses a customer’s Gainsight configuration export (a 25-megabyte Excel file spanning 14 structured tabs) and evaluates it for structural signals like:

• Stale scorecards

• Journey Orchestrator programs that are limited to email sends

• Playbooks that exist but never fire

• Rules that have not triggered in over a year

Value Narrative, which takes those technical findings and translates them into business consequences. For example, if Config Intelligence detects a stale scorecard, Value Narrative surfaces the risks that issue poses. In this case, a stale scorecard could lead to a CSM missing warning signs.

Together, these capabilities allowed the system to generate an end-to-end Instance Review deck that looked familiar to a CSM, but was produced in seconds rather than hours. And since each of these capabilities provided a standalone workflow, not only were we tackling IRs, we set ourselves ahead for future projects.

The Big Test

Once the prototype worked, we tested it on data from our own instance and compared its results to a human-written Instance Review. This gave us a direct, side-by-side comparison.

The AI found many of the same issues as humans, such as:

• Low dashboard use

• Underutilized Success Plans

• Limited Journey Orchestrator engagement

• Unused reports

• Outdated scorecards

It then created a review deck in seconds that would have taken much longer to build by hand. Most importantly, it brought consistency: the same data always led to the same findings, eliminating differences caused by manual review.

The overall match rate compared to human-led IRs landed around 60–70%, which turned out to be a clear win. CSMs weren’t starting from zero anymore and could focus on applying judgment rather than assembling slides.

What the AI Couldn’t See

We weren’t at 100% parity yet, but this was an opportunity to improve and learn more about the areas where results overlapped.

Here’s what we discovered:

The AI could not recognize strengths. It only pointed out problems and couldn’t say, “This is working, don’t break it,” as human reviewers often do. This gap matters because it shapes the conversation itself. A review that focuses on problems can derail what should be a constructive discussion. So, we have already started adding strength checks to ensure our reviews are balanced.

The AI’s strategic recommendations were also shallow. Its suggested approaches were broad, like “consider training.” Meanwhile, a human could suggest a deeper approach, such as establishing cross-functional ownership for digital work or tying health scores to measurable results. These suggestions were on the right track, but not very useful in practice. This simply means we need to keep massaging and fine-tuning our prompts and n-shot examples.

Even more telling were the gaps we didn’t expect. Both of our human reviewers identified unused AI features (i.e., Copilot, AI Takeaways, and AI Follow-Up) as clear opportunities, but the prototype missed them because those details were not included in the export. This wasn’t a problem with the model, but a limitation of the data: you can’t derive insights from information you don’t have.

Going Back to the Drawing Board

One particular discrepancy ultimately reshaped the prototype’s design. Both human reviewers said the customer’s usage of Gainsight’s Timeline feature was healthy and high, but the AI marked it as underused at 16.9% adoption, which was below its set threshold.

At first, it looked like the same data led to different conclusions. But when we looked closer, we realized the data wasn’t actually the same at all.

The problem was context. The AI couldn’t see what CSMs know from regular customer conversations or the details that affect how features are used. Maybe the customer isn’t using a particular feature due to security policies or other company constraints. Or perhaps the CSM is using relative judgment based on the company’s usage of other features. Meanwhile, the AI is using a strict rule.

This showed that our system was missing key information that only lives in a CSM’s head. It’s an important reminder that keeping humans involved is part of the design, not a limitation.

When the Users Pushed Back

We demoed the prototype to a group of CSMs and ops leaders who would actually be using it and received immediate (and humbling) feedback.

One CSM looked at the output and said, “An exec would look at this and say ‘so what?’ We have to translate these findings into the why.”

She was right. We built a strong tool for CS ops, with deep diagnostics, precise metrics, and detailed recommendations, but we missed that this is not what executives need. A VP of Customer Success doesn’t want that level of detail. They want the story. What does this mean for my business? What should I prioritize? What’s the impact?

The group made the division of work clear by letting the AI handle detailed analysis while humans added judgment, context, and storytelling.

What the Prototype Revealed Beyond the Plan

The feedback sessions revealed what we had misunderstood, helping de-risk the decision to move forward.

We proved we could parse complex exports, apply human-like rules, and generate coherent output. These assumptions became facts only after the system was tested on real data.

We also found problems that planning alone would not have revealed, such as missing AI feature adoption in exports, viewer licenses that make usage numbers look higher, and thresholds based on outdated documentation. These are real, common issues, and investing in production without learning this would have been a risky guess.

If you take away anything from this article: don’t rush straight to production. We could have spent months refining requirements for a production build and still missed the Timeline problem, the export gaps, and the “so what?” feedback. The prototype surfaced all of that in three weeks.

The Lessons That Stuck

1. Speed is how learning happens.

We spent one week understanding the problem and two weeks building, so by week three, we had something real people could critique. Method Garage helped us move fast, and that tempo mattered. If we had spent months refining requirements or debating architecture, we’d still be arguing over schemas rather than learning from real output.

2. “Human-in-the-loop” as a key feature.

CSMs have customer knowledge and context that an Excel sheet can’t replicate. There’s so much push in the AI industry to do everything and take action. But AI plus humans is where the real magic happens. Humans aren’t just there to do final review before sending. They’re part of the intelligence and intentionally part of the AI-enabled workflow.

3. You don’t get the design right without your users in the room.

The demo session with CSMs and ops leaders was more productive than any planning meeting we could have run, because people were reacting to something tangible. They could see what worked, what didn’t, and how to improve it.

4. Prototypes reduce risk by exposing failure.

Even when it wasn’t polished, the prototype showed what was possible and what wasn’t, helping us decide how to move forward. That’s why prototyping isn’t just a phase before the real work. It is the work, and often the only way to find out which assumptions deserve to survive.

What’s Next

At 60–70% match with human reviews, the prototype gave us what we needed: clarity on what’s worth automating and what still needs a human.

Now we’re fixing what we found. CSMs are rewriting thresholds based on how they actually evaluate usage so we can fix the Timeline misdiagnosis. We’re building two separate outputs to solve the “so what” problem: detailed diagnostics for CS Ops and business narratives for CS Execs.

This is the reward for doing discovery well. You build a system that keeps getting smarter as you use it for new problems.

Part 1 was about learning before building. Part 2 was about what we learned once we started building and left the safety net behind.

Both matter. Neither is optional.

Method Garage played a huge role in getting this off the ground. Their strength is translating messy, human workflows into something structured enough to build against. They surface the knowledge most teams skip and force the hard decisions early. If you’re applying AI to complex, judgment-heavy processes, that kind of rigor matters.

If you’re experimenting with AI within Customer Success, especially in agentic, workflow-heavy environments, I want to hear what’s actually holding up under pressure. Drop a comment or reach out directly.

👋 Connect with me on LinkedIn

🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

What Consumption-Based CS Looks Like Inside a $100B Company ft. Jared Collins (This Week’s [Un]Churned 🎙️)

CSMs turned data strategists. Consumption-based pricing models. Proactive playbooks. This week on [Un]Churned, Jared Collins, Senior Director of CS at Dell Technologies, deconstructed this jargon salad and transformed it into practical advice for CS teams, big and small.

“CSMs don’t work in a vacuum. They’re not going to be successful on their own…If the AE is the CEO of the customer relationship, the CSM is the COO.” A quote from Jared that rings especially true, considering his Dell Apex team is part of the massive broader whole of Dell Technologies.

He shares how his team focuses on building strong internal and external relationships, and uses data, triggers, and simple playbooks to connect the dots and stay ahead of customer risk.

With over 16 years at Dell, a foundational player in the infrastructure and tech industry, Jared’s perspective on the future of software is one that can’t be ignored. I’ll give you a hint: zooming out on goals and zooming in on technical CSM skillsets are both part of the path ahead.

“Go Win. Go Carry the Day”

This was definitely a ‘pinch me I’m dreaming’ moment. I had the privilege to (virtually) sit down with Robert F. Smith, Chairman and CEO of Vista Equity Partners, for a company-wide fireside chat about the future of SaaS, his leadership philosophy, and how his commitment to building access to equity in marginalized communities.

He left us with these parting words: “Go win, go carry the day. Win everything, make it happen. Have a great time. This is an amazing time in life, and you guys are at the forefront. Not only keep the lead, but literally go capture it all, because it’s yours to capture.”

Check out the full post, not just for my takeaways, but to see what resonated with the rest of the Gainsight team.

Read the full post →

Translating CS for the Boardroom

If you’re headed into your next board meeting with a story about how Customer Success “saved the day”, but have no numbers to back it up, you’re better off declining the cal invite.

It’s a harsh truth, but one that Teresa Anania, CCO at Sophos, provides the cure for in her article, “The Attribution Model That Survives Board Scrutiny.” Drawing on her experience at Autodesk, Zendesk, and now Sophos, Teresa’s developed an attribution model that translates CS impact into financial credibility.

This article is a deep dive into her process, the constraints she encountered, and how you can learn from her experience to build an attribution model of your own.

Read the full article →

Evolving to Agent-Led Growth

Mickey Alon, co-founder and CEO of Foldspace AI warns, “The tolerance for "learning curves" has hit zero.” Forcing users to memorize workflows is old news. Giving them an agent to handle execution is what’s next.

If you’re exploring agentic use cases for your product, I highly recommend giving this a read.

Read the full post →

Bonus [Un]Churned Episode: 75% of SaaS Companies Will Disappear ft. Brett Queener (Bonfire Ventures)

“It’s irresponsible to not understand what’s possible.” That’s the advice venture capitalist Brett Queener shares on this bonus [Un]Churned episode. He also challenges his past perspective on AI and shares his unfiltered thoughts about the direction the industry is headed in.

His take? Judgement is now the biggest moat (and he’s written all about it in his recent Substack article). Queue up this bonus [Un]Churned episode the next time you have 16 minutes. Otherwise, you’re just being irresponsible.

Wrapping Up

My mind’s still swimming with all the forward-looking conversations I’ve had this week. All I can say is, the bar continues to rise, and we can’t win by playing defense. What are you doing to meet it?

See you next Monday 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Answering this question requires CS leaders to move beyond storytelling and clearly define the tangible outcomes that would be lost if their team did not exist. This clarity has become increasingly critical as companies scale, margins tighten, and post-sales investments must compete with priorities like security research and product development.

If you keep getting asked to defend Customer Success with stories, you’ll soon find that anecdotal stories just don’t survive in a room full of engineers.

Teresa Anania, Chief Customer Officer at Sophos, addressed this by narrowing in on the Customer Success narrative. Throughout her time at Autodesk, Zendesk, and now Sophos, she developed and iterated on attribution models that show where, when, and under what conditions Customer Success drives measurable outcomes.

What follows is how her attribution model was built, along with the constraints that led executives to trust the results.

Why Most CS Metrics Fail When Money Is on the Line

When presenting to boards, ambiguity undermines credibility. Metrics such as Net Promoter Score (NPS), health scores, and activity volume only show what happened, not what changed as a result of Customer Success.

To create a narrative that definitively proves CS’s impact, you have to go against your instincts. Teresa’s first decision was philosophical: attribution would only cover work that could plausibly alter renewal or expansion behavior. Other activities, while operationally important, would not be included in the analysis.

That decision alone made the model narrower and more defensible.

Step 1: Decide What Work Is Allowed to Matter

Before starting analysis, Teresa’s team had to define which Customer Success engagements were eligible for attribution.

To do that, the team built prescriptive playbooks by segment. These playbooks laid out which actions should happen at key points in a customer’s journey.

For Enterprise and Strategic accounts:

• Welcome Playbook: Structured onboarding with clear milestones

• Quarterly Business Reviews: Templated sessions focused on proof of value delivered—not “how’s it going” calls

• Security Posture Reviews: Technical deep dives into the customer’s security footprint

• Roadmap Reviews: Product alignment sessions and forward-looking demos

• Executive Connect Program: Executive sponsor engagement tied to strategic accounts

For Tech Touch segments:

• Onboarding campaigns: Webinar-based programs

• Adoption intervention campaigns: Triggered by product usage signals

• Renewal preparation campaigns: Webinar plus targeted 1:1 outreach for key accounts

Notably excluded were relationship check-ins, reactive escalations, and ad hoc conversations. While interactions mattered, they could not claim renewal impact by association.

It’s at this step where models often break. Teams naturally want attribution to validate their efforts. Teresa addressed this by designing an attribution model to check what really moved the needle.

Step 2: Build Cohorts You’d Defend in a Boardroom

With eligible activities defined, the next challenge was comparison. Attribution only works if you’re comparing customers who received help versus those who didn’t. For each activity type and segment, customers were grouped into three cohorts:

• Unengaged look-alikes, matched on segment, ARR, tenure, geography, and baseline health

• Engaged accounts, which received the defined engagement within a six-month window

• Highly engaged accounts, which received the engagement at or above the intended cadence

To avoid rewarding last-minute saves or recency bias, outcomes weren’t measured immediately. Renewal and adoption results were evaluated on a rolling six months post-engagement window, adding credibility to the correlation as it takes time to influence renewal outcomes. If you look too close to the renewal, you’re just measuring heroics. Teresa understood that her team needed distance so the signal could show up.

Step 3: Clarify What the Data Shows

Once the cohorts and look-back window were set, the data started doing its real job: narrowing what Customer Success could credibly claim as impact.

For high-touch and strategic accounts, those who received the Customer Success Manager (CSM) high-value engagement renewed more often than similar customers who didn’t. The CSMs who adhered to the full engagement touchpoint plan did even better. The exact numbers were less important than the trend. Some efforts actually changed customer outcomes; others mostly felt meaningful but didn’t move the needle.

The data kept pointing to a few simple truths:

• Structured, outcome-driven engagements mattered. Customers who received them renewed at higher rates than similar, unengaged accounts.

• Consistency amplified impact. Accounts that received the full engagement plan outperformed those with less frequent touch.

• Not all playbooks earned their keep. Some activities showed little or no correlation with renewal and lost their claim to commercial impact.

• Tech-touch worked selectively. Targeted intervention campaigns outperformed broad onboarding or webinar programs, which still helped but delivered smaller, more predictable lift.

Once Teresa and her team clarified what was and wasn’t working, they were a step closer to answering what happens when CS isn’t there. Realistically, this doesn’t show up as removing CS altogether but as changes in coverage. What would happen if there were fewer accounts touched or entire segments without an assigned CSM?

To answer this question, they shifted their focus to understanding how much coverage it takes to see results and how much risk is incurred if coverage changes.

Step 4: Translate Engagement Into Coverage Economics

Attribution only becomes useful when you can model what happens as coverage goes up or down. The math itself was intentionally simple:

• Coverage percentage based on CSM capacity and portfolio size

• Total Book of Business (BoB) ARR

• Observed renewal uplift attributable to engagement

Multiplying those inputs produced a dollar figure tied to coverage. That figure could then be adjusted to simulate changes, such as adding headcount, reducing coverage, or shifting certain customer segments to tech-touch only.

Once this model existed, headcount discussions can take on a different tone. Optimizing CS spend stopped being about efficiency and became a trade-off discussion about renewal risk. Executives may still make tough resourcing decisions, but the trade-off risk can be measurable and helpful in linking those decisions to Company retention targets.

How the Model Actually Gets Used

Once activity attribution was in place, it stopped living in dashboards and started showing up in decisions that had previously relied on industry benchmarks or opinion. Here are three situations in which Teresa was able to apply this attribution:

1. Capacity Planning Linked to Coverage Impact

Instead of starting with industry ratios, Teresa’s team flipped the script and started with coverage. The attribution model can help determine the optimal segmentation strategy where you get the biggest bang for your buck in human coverage across your BoB:

• Where does it make the most sense to assure High-Touch coverage and what is the GRR and NRR impact of that?

• The historical uplift tied to those engagements.

• How much ARR was coming up for renewal to quantify impact by quarter.

Sometimes, optimizing coverage (i.e., widening ratios, scaling CS to tech touch, etc.) looked smart on paper, but the renewal risk was way higher than the savings. Other times, the model can help determine where it isn’t worth adding more CSMs when tech touch can be a way to scale. The model can make both calls clear.

2. CSM Capacity Becomes Easier to Evaluate

Portfolio sizing debates also shifted. Rather than using only industry benchmarks, the team looked at where attribution impact began to degrade as account loads increased. That point became the boundary, once outcomes flattened and then declined.

This way of thinking reframed capacity discussions. Instead of asking more from CSMs, the strategy shifted to identifying where the work no longer produced accretive results.

3. Performance Conversations Happened Sooner

Attribution changed how the team handled performance by surfacing risks well before renewals. If a portfolio didn’t do well, the team looked at which actions didn’t happen or didn’t work, rather than just explaining the numbers after the fact.

It became easier to spot the team’s strong performers. CSMs who consistently executed the prescriptive playbooks showed measurable uplift compared to similar portfolios, making it harder to dismiss the differences as luck or account quality.

What made this workable was consistent logging, a restrained scope, and accepting that some work would never be attributed. The model allowed Teresa and her team to identify, reward, and replicate what actually changed outcomes.

Where AI Helps (and Where It Doesn’t)

AI only came into play after the CS impact attribution model had already set some guardrails. The order mattered—a lot. If you don’t know which actions actually move the needle, AI just gets you doing more stuff, not better stuff.

When the timing was right, AI turned into a force multiplier. Suddenly, Teresa’s team could do a few things faster and more reliably than if they were flying solo:

• Refine segmentation by identifying which customers benefited most from which playbooks, based on observed attribution lift rather than assumed value.

• Adjust engagement cadence by showing where additional touchpoints produced diminishing returns, making it easier to scale back human effort without losing impact.

• Scale tech-touch execution in segments where attribution showed marginal human lift, allowing automation to replace work that was already proven to be low leverage.

What AI did not do was create causality. It didn’t tell the team what mattered. It only helped them act on what the attribution model had already made clear.

That’s an easy (and risky) line to blur. Without attribution, AI just speeds up uncertainty. It makes everyone busy, but mostly just doubles down on old assumptions. When attribution comes first, though, AI amplifies what already works and helps cut out the noise.

That’s why Teresa used AI as a finishing touch, not the centerpiece. The model had to prove itself before it got automated.

What This Model Ultimately Forces

Creating a CS impact attribution model can cause a ripple effect that goes beyond technical impact to an organizational one. Once the model exists, Customer Success loses the ability to rely solely on narrative. Low-impact work, performance variances, and coverage risks are out there for everyone to see. Meanwhile, other efforts are shifted out of focus if they’re not directly impacting retention or revenue.

As Teresa put it during our conversation, “You have to be okay knowing some of the work will never show up. If you try to force it, you lose credibility.”

That discomfort is precisely why the model holds up under scrutiny. The true value comes from proving CS is a revenue driver. Once that argument is clear, leadership can move forward with what they’re willing to invest in and what they’re willing to give.

About This Article

This article draws from a detailed conversation with Teresa Anania, Chief Customer Officer at Sophos, on how to build Customer Success attribution models that hold up under executive and board-level scrutiny. Teresa has refined this methodology across multiple organizations, including Zendesk and Autodesk, adapting it to different customer segments, operating models, and growth stages. All data referenced here has been sanitized to protect confidential business information.

Here’s what’s been on my mind this week.

A Masterclass in Category Creation from the Leaders at Gong and Gainsight ft. Udi & Lauren (This Week’s [Un]Churned 🎙️)

This week on [Un]Churned, we interrupt your regularly scheduled programming for a refreshing episode that looks back on how two marketing heavy hitters created entire categories for their products.

This episode isn’t just for marketers. It’s for anyone that wants to see the real work that goes into bringing a solution to a market that needs it, but doesn’t know it yet. Udi Ledergor, Chief Evangelist and former CMO at Gong, and Lauren Olerich, Sr. Director of Corporate Marketing at Gainsight, rewind time and talk about how they first established product market fit and what it takes to discover the voice that really hits with your audience.

They share the key moments that brought Gong and Gainsight to where they are today, like prioritizing community, educating an entire potential market, and using big fonts to get big ideas across (IYKYK).

The next step? Disrupting the very categories they’ve created as their companies grow to fit the AI boom. This is a great listen for anyone ideating on or launching something new this year. Lauren and Udi’s experiences are guaranteed to get you in a ‘break-the-mold’ mindset.

“Claude was the hands. I was the head.”

What do Ratatouille, The Matrix, and software have in common? They were all inspiration for our CEO, Chuck Ganapathi’s, recent vibe code project: ‘Glue’, a mobile native app he uses to prepare for customer meetings. In this article, Chuck reflects on just how quickly the software industry has changed.

When he was at Siebel, nine months to launch a new product had him “amazed at how such a small team could drive so much innovation in such a short period of time.” But here’s the kicker: Glue went from idea to reality in five days.

Read the full post →

Not Churning ≠ Growth

We’re all feeling it—SaaS is figuring out how to navigate the AI disruption in real time. As Jason Lemkin puts it, “I am less and less sure every single week what B2B software looks like in 2027 and beyond.”

In this article, Jason works backwards to see how we got here and looks at the opportunities that lie ahead.

If we’re going to succeed, we need to own the AI agents that make our platforms thrive. Otherwise, we’ll find ourselves in the low-churn-flat-growth purgatory.

It’s a grounding reminder to not panic, move fast, and focus on innovation that will keep you relevant.

Read the full post →

News From the AI Epicenter

How do you talk to your friends and family about AI who aren’t in the tech industry? I ran into a neighbor that shared this viral article with me and I haven’t been able to stop thinking about it.

Matt Shumer’s post is a sobering and realistic look at just how insanely fast AI has advanced in the last few years—heck, even the last few months.

We’ve seen it firsthand. AI is going beyond busywork and handling true cognitive judgement and intuition, impacting jobs that were regarded as “safe” for at least the next few years. No matter what industry you work in, the knowledge gap is closing quickly and the time for early adoption has passed.

Read the full post →

Bonus [Un]Churned Episode: How Do You Measure AI Proficiency in Your CS Team? ft. Guy Galon (Obrela)

This 16-minute [Un]Churned episode is packed with insights from Guy Galon, CCO at Obrela. He shares how curiosity, flexibility, and thinking ahead gives you an advantage when it comes to customer-facing roles and building AI strategies. If you’re interested in learning how to build a multifaceted CS team and where AI comes in, hit play and get inspired.

Wrapping Up

Sometimes you have to take a step back to see the full picture. While the industry might be rapidly changing, the lessons you’ve learned over time still matter. These are discussions worth having with the people in and out of the tech industry because the call to action is now.

See you next Monday 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Here’s the letter (with minor edits to remove confidential info) I sent my team at Gainsight.

January 25, 2026

Gainsters,

It was Christmas of 1999 and I was a freshly-minted MBA working as a product manager at Siebel Systems, the fastest growing technology company in the United States with 782,978% growth over the prior five years. Sitting near me was a whip-smart CS grad from Cornell, Chew Lam. He and I were under tremendous pressure to release eChannel, Siebel’s first-ever web based product for managing channel partners. (Every era has its naming trend: the ‘e’ prefix then, the ‘i’ prefix in the mid-2010s, the ‘AI’ suffix now.)

You see, Siebel was a client-server platform with a “thick client” built on Windows (aka desktop apps today). Everyone in Silicon Valley was talking about how the internet was going to disrupt client-server software, so Siebel was under the gun to demonstrate that we knew how to build web apps. Chew and I felt the weight of that responsibility, along with our teammates in the Internet Apps team in R&D.

We toiled for months to build eChannel: back then a full release could take about a year. I was churning out voluminous MRDs and Chew and the other engineers on the team were pushing lots of code. Finally, after 9 months of working late nights, weekends, and holidays, our team finally launched eChannel as a part of the Siebel 6 release in April 2000. eChannel became the fastest growing product at Siebel, to reach $100M in revenue. Having come from the consulting industry, I was amazed at how such a small team could drive so much innovation in such a short period of time.

Fast forward to 2025. Chew and I were reunited at Gainsight. But this time we were dealing with the Salesforce security advisory and an escalation at one of our large enterprise customers on Christmas eve. When things quieted down and I reemerged after holiday over-indulgence, I noticed a post on X that changed my outlook forever.

Boris Cherny, the creator of Claude Code at Anthropic, casually called out that 100% of the code he wrote in the last 30 days to enhance Claude Code was written by Claude Code. One. Hundred. Percent.

To me, this was crossing the Rubicon.

Everything I had known about software development was turned upside down. If Claude Code is your coder, your only limits are compute, imagination, and experience.

I admit I got Claude-pilled.

I went on a Claude-bender to build something I needed for myself personally since the time I joined Gainsight but had never made the cut on our product priority list. Over the course of two weekends – five days total, thanks to MLK Day – and in-between driving my kids around for their sports activities, I went into full builder mode.

I had not programmed in over 25 years. In fact, I’d only written code in grad school, never for commercial software development. I opened up the Claude desktop app and started with a simple prompt:

Claude AI probed me about the purpose of the app and its target users, discussed some architectural options and tradeoffs, designed the UI, architected the app to be secure end-to-end, taught me how to use Claude Code to build it, and gave me the prompts to enter into Claude Code’s command line interface (CLI).

Claude Code wrote the code, pushed to a Git repository, and deployed to production, and walked me through how to use Apple App Store Connect and TestFlight to get the app on my phone.

Roughly 820 conversation turns, 40,000 lines of code, and 18M tokens later, Glue was born. Glue is a native mobile app that combines the best of Gainsight and Staircase, with a little Claude magic thrown in. It gives me customer intelligence at my fingertips on the device I use most: my iPhone.

Why is this app important to me? My days are filled with meetings and I always feel unprepared walking in. So here is what I try do as fast as possible:

Open my laptop.

Go to Chrome.

Launch Gainsight.

Review the C360 screen and the timeline.

Switch to the Staircase AI tab and read the AI summary and health score

Back and forth conversation with Ask Staircase.

Look up the person I’m going to meet and try to figure out their sentiment.

Jump to LinkedIn.

Search for their name.

Review their profile to see what icebreakers I can use.

Glue changes all that right from my phone. Here is what’s special about it:

• Aggregates information from Gainsight, Staircase, and LinkedIn in a way that gives you what you need at a glance.

• Natural language interface (Ask Glue) that brings together unstructured insights from Staircase and structured data from Gainsight into a single Claude-generated answer.

• Native mobile app with local storage for fast loading and Bloomberg-Terminal-inspired UI. (Personally, I think Bloomberg Terminal’s information design and aesthetic is one of the best examples of a serious tool for serious people who need data fast.)

• Security-first design.

Here is a preview of the app:

In addition to the app itself, I asked it to create some artifacts to explain what we had built together:

• Feature documentation, for end users to see what the app does.

• Project spec, for engineers looking to understand how the app was built.

• Architecture spec, for architects to understand the design principles.

• Security assessment, for our CISO team to check if the app meets guidelines.

• Screenshots, for our product marketing team to use in collateral.

I also asked Claude to estimate how long it would take the typical R&D team to build the app without AI.

The answer? A team consisting of a product manager, designer, front end engineer, back end engineer, iOS engineer, and product security engineer, would have taken 3-4 months to build Glue.

Claude and I built this in 5 days.

To be clear, I had some help. Our IT team created an Okta app to enable Glue’s authentication, the Staircase team gave me an end point to call their MCP server, and our Admin team helped set up a Rest API connector to our internal Gainsight-on-Gainsight instance.

I also have the benefit of being in the software industry since 1999. I’ve had the benefit of learning from world-class developers like Chew and countless others across Siebel, Salesforce, and Gainsight. Having worked in CRM for most of my career, I understand customer data models. My startup, Tact.ai, was all about mobile and AI CRM for on-the-go professionals. So, yes, I have the benefit of those lived experiences.

Still, it is absolutely bonkers that someone like me could build something as sophisticated as Glue. Now, there is no turning back.

One of my favorite movies of all time is Ratatouille, where a rat chef was inspired by Chef Gusteau’s book and TV show “Anyone Can Cook.” Now, with Claude Code or any of the other vibecoding platforms, Anyone Can Build.

My mind is racing with all the ways each one of you could leverage Claude Code to build apps that solve specific problems you face in your day to day work. I can imagine an AE building their own battle card generator; an HR business partner building a manager coaching assistant; an accountant building an expense report auditor; or a RevOps analyst building a whitespace detector. The democratization of software creation isn’t a fantasy, it’s happening now.

Which raises an uncomfortable question for those of us who sell software for a living: If everyone can build, why would anyone buy?

One of my favorite UX designers, Luke Wroblewski, posted the following on LinkedIn a few days ago:

Josh Schachter pointed me to another interesting post by CS executive turned investor, John Gleason, about companies building their own CS platforms:

I’ve lived through every major platform shift since 1999. Client-server to web. Web to cloud. Cloud to social and mobile. Each one changed something important…where software ran, how it was delivered and monetized, when and with whom you could use it. But the thing that never changed: professional engineers wrote the code, companies sold it, customers bought it.

AI breaks that pattern. It doesn’t just change where, how, or when. It changes who. The person with the problem can now build the solution.

The moat isn’t code anymore. It’s context. It’s domain expertise. It’s knowing what a frantic executive needs in the two minutes before a customer meeting.

Claude was the hands. I was the head.

And that’s exactly how I think about Gainsight’s position in this new world. We’ve spent 15 years obsessing over customer success. We’ve got playbooks and proprietary expertise aggregated from thousands of companies. Our R&D team is already building with agentic AI. Almost 90% of Atlas code is written by AI.

A customer can absolutely spin up their own CS tool with Claude. But can they match 15 years of domain expertise? Can they iterate as fast as a team that’s done nothing but this? I don’t think so.

The winners will not be the ones who write the most code. They’ll be the ones who understand the problem best.

We know retention. That’s our head. AI is just the hands.

So what are you going to build?

Cheers,

Chuck

Here’s what’s been on my mind this week.

What Happens When You Apply Digital Motions to Your Biggest Accounts? ft. Christine Lavery (This Week’s [Un]Churned 🎙️)

This week on [Un]Churned, Christine Lavery, VP of Customer Success at Conga, shared a principle that I’ve never heard in all 150+ episodes of the podcast: digital CS should serve the entire customer ecosystem, not just the long-tail.

As Christine puts it, “At the end of the day, we want our users to use our product and maximize how they’re using it. That is the same no matter the size and span that you have with us.” If you’re reading this and thinking, “no shit”, I highly recommend you press play on this week’s episode.

Christine’s diverse background includes managing both white-glove enterprise CS at Typeform, and long-tail, PLG-led motions at Agent Sync. It’s clear how these experiences primed her to become the perfect strategist for this ‘1:all’ strategic motion. You’ll get her entire playbook, with tactics around customer journey mapping, achieving buy-in at the team and executive level, and collaboration with Marketing, Supporting, and PMM.

We also talk shop and hit on CS hot topics like AI’s role in CS and how living in the revenue org has opened up opportunities for expansion-driving partnerships.

If you’re interested in building a ‘1:all’ motion, Christine’s playbook is a must-have. If you’re not convinced, give us 34 minutes to change your mind.

The Hidden Cost of Building Fast

Prem’s post got me thinking about how we really took the limits imposed by engineering for granted. When building took time and money, it forced us to make choices. With those choices came focus. And with that focus we had intentional roadmaps.

Without that guardrail, teams can build faster than customers can absorb. This topic isn’t new, especially on [Un]Churned, but I think it’s important to come at it from all angles. Prem’s post isn’t about Customer Success directly, but it points at a shift CS leaders are already feeling.

If you’re seeing this tension show up with your customers, it’s worth reading.

Read the full post →

If You Build It… They’re Probably Building It Too

If you ever need a reminder of how fast the industry is moving, scroll LinkedIn. Just days after sharing my thoughts on build vs. buy for a Customer Success Platform, a post showed up from Ralphie at Deepgram sharing how her team built their own; and how much they improved it in one month.

This is the reality we have to lean into. Teams are building tools for themselves. And honestly, it’s impressive. When you know your own problems really well, you can move fast and make something that fits exactly what you need.

But building the first version isn’t the hardest part. The real challenge is keeping it running as the company grows. Launching, iterating, and maintaining a CSP at scale is a Herculean task.

That’s why teams that build often become the best buyers later. Building shows you where things get complicated and costly over time. It doesn’t make you anti-vendor, it actually helps you understand what the work really requires as you grow.

Read the full post →

Wrapping Up

Digital CS scales because it has to. Building shows up because teams want control. The leaders who win are the ones who treat both as real choices, not side projects.

See you next Monday 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

This CCO Won’t Ask His Team To Do What He Wouldn’t Do Himself ft. Jim Richmond (This Week’s [Un]Churned 🎙️)

This week’s [Un]Churned episode touched on something simple that’s quietly getting harder to practice: staying close to the work.

As things speed up, it’s easy to spend more time planning and building frameworks. That can feel like good leadership, but it also takes you further from the pressure your team feels. This distance can change your decisions without you noticing.

In this week’s episode, Jim Richmond, Chief Customer Officer at Smartling, talks about holding himself to the same standard he expects from his team. Knowing where something breaks because you’ve felt it break. Knowing when a push is reasonable because you’ve carried it yourself.

We also talk about how he uses AI to clear out the busy work so he can stay close to the parts of the job that require judgment. When the routine work disappears, what’s left are the kinds of interactions that actually impact retention.

Falling Behind, or Just Feeling It?

I shared a post this week because the questions felt more honest than any answer I was ready to give.

Recently, a few people I trust mentioned building things in hours that used to take weeks. They described rough versions being built and put into use almost immediately. I expected that to feel exciting. Instead, it left me uneasy.

I’ve spent years building in this space and watching cycles come and go, yet this moment feels different. The pace is faster, but what’s changed more is the expectation around speed. Decisions that once felt careful now feel slow, and that shift creates a kind of pressure I’m still trying to understand.

I don’t yet know what the right posture is, but pretending it isn’t there feels worse.

→ Read the post here, and let me know: does this resonate, or am I just behind?

Early Experiments, Better Decisions

This article from 3Sixty Insights describes a moment many teams now find themselves in before choosing a long-term solution. They build small tools with AI, test them in real-world use, and learn what actually matters. It’s less about speed and more about clarity. By building first, teams start to see the real problem they’re trying to solve.

Over time, these tools often reach their limits and become difficult to manage or scale. This process helps teams clarify their needs and pinpoint areas where they require support. If you’re interested in seeing an example of how early experimentation can lead to confident and well-prepared teams when scaling solutions.

→ Read the full post

Wrapping Up

The same pressure that creates anxiety for leaders is driving teams to experiment on their own. Faster cycles reward action, rough learning, and early signals. What matters is how those signals get absorbed, and whether they lead to clearer decisions or quiet workarounds.

See you next Monday 🧠

🌱 Josh

SVP, Strategy & Market Development @ Gainsight

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

(This Week’s [Un]Churned 🎙️)

Episode 164: Why Your Post-Sales Budget Should Be 7% of Revenue (Not 10%) ft. Abbas Haider Ali

I had Abbas Haider Ali, SVP of Customer Success at GitHub, on [Un]Churned this week to talk about post-sales investment, proving ROI, and where AI actually fits into all of it once the hype fades.

Abbas made a distinction that really stuck with me. Customer Success isn’t underfunded nearly as often as it’s misallocated. When teams default to adding headcount or inflating coverage models without tightening ownership and outcomes, spend goes up but impact doesn’t. His case for anchoring post-sales investment closer to 7% of revenue anchors on discipline and intent.

We dug into what breaks when CS budgets drift and how to think about efficiency without devaluing the work. If you’re trying to defend CS investment, rethink your operating model, or prepare your team for an AI-shaped future, this episode is a practical place to start.

Stay tuned for Abbas’ complete playbook. He’ll share thoughts and tactics around resourcing, AI leverage, and building a defensible post-sales model that holds up under scrutiny.

Horizontal SaaS Meets the CS Reality Check

This week, Luke Sophinos laid out a blunt argument in his newsletter: horizontal SaaS is heading toward an extinction event as AI agents take over more generic workflows.

What survives won’t be tools built for everyone, it’ll be systems that own something hard—deep context, proprietary data, durable workflows, and real accountability. This isn’t about interfaces going away. It’s about where value actually lives once execution becomes automated.

From a Customer Success perspective, this hits close to home. The platforms and teams that win will go beyond surfacing insights to actually helping people act, decide, and prove outcomes. Generic tooling and shallow signals won’t hold up for long. As the bar rises, CS leaders will feel that pressure early (if not already).

Read the full post →

The Claude Constitution (Through a Customer Success Lens)

I spent some time this week reading Anthropic’s Claude Constitution, and while it’s positioned as a framework for building safer, more aligned AI systems, I couldn’t help but read it through a Customer Success lens. As Anthropic describes it, the Constitution defines the kind of entity Claude is meant to be and the values that should shape its behavior.

That framing feels increasingly relevant for Customer Success. Training AI agents isn’t just about automation or efficiency. It’s about judgment. When an agent flags risk, recommends action, or moderates a customer interaction, it’s making decisions that affect trust. Those decisions need context, boundaries, and an understanding of consequences, not just data.

As AI agents take on more responsibility in post-sales, this is the lens CS leaders should aim to use. Agents should act in ways that are explainable, defensible, and aligned with real customer relationships. Values and guardrails can’t be layered on after the fact. They have to be part of how agents are trained from the start.

That’s what makes the Claude Constitution a useful reference point. Not as a policy document, but as a reminder that in Customer Success, alignment isn’t optional, it’s foundational.

Read the full constitution →

Wrapping Up

Automation has a way of removing the excuses. As AI starts doing more of the execution, only the elements of a business built on judgment, clarity, and accountability really hold up. Bigger teams and louder tools stop being the answer pretty quickly.

The more interesting question is what the work looks like once the training wheels come off. How are you designing it? I’d love to hear.

See you next Monday 🧠

🌱 Josh

SVP, Strategy & Market Development @ Gainsight

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

(This Week’s [Un]Churned 🎙️)

Episode 163: 18 Months, 7,000 Customers, 67% Support Resolution: Inside Intercom’s FDE Strategy ft. Diego Ballona

This week on [Un]Churned, I sat down with Diego Ballona, who leads Intercom’s Forward Deployed Engineering team. In 18 months, his group built a system that now resolves 67% of support questions before they ever reach a human, across roughly 7,000 customers.

While the stat is impressive, it’s not even the most interesting part.

What made this conversation worth lingering on was how deliberately unglamorous the work was. Diego didn’t describe a breakthrough moment or a big-bang AI rollout. He described a series of choices about where not to intervene, which questions weren’t worth automating yet, and how much ambiguity the team was willing to tolerate while learning what customers actually needed at the front door.

A lot of support leaders talk about being customer-centric. Diego talked about being sequence-centric. What information should exist before a customer asks for help? What should be answered without context? Where does automation help, and where does it just move the problem downstream?

We also spent time on a tension that most execs feel but rarely name clearly: efficiency versus trust. Diego’s take was refreshingly pragmatic: automation isn’t there to replace humans; it’s there to earn the right not to involve them. When that framing is clear, the trade-offs become easier to make and easier to defend internally.

Think of this episode as a field report from a team that treated support like a product surface, not a cost center. If you’re scaling a company and wondering how much of your support load should exist before a ticket is ever filed, this is a model worth studying.

Building in Public (and Being Wrong First)

Earlier this week, I shared a post on that was a little more candid than usual. Not because it was provocative, but because it admitted something that’s uncomfortable to say as a leader: I thought we understood our own Customer Success process. We didn’t.

That post was a preview of a deeper piece we just published with Saul Gurdus at Method Garage, where we turned the lens inward on one of the most valuable yet avoided CSM motions: the Instance Review. What I assumed was an AI problem turned out to be something else entirely: a capability gap hidden behind years of workarounds and tribal knowledge.

In the article, I walk through what actually mattered in the Blueprint phase. Finding the one person who had already cracked the problem manually. Letting go of the idea that we needed perfect data to start. And realizing that the real ROI of AI is finally being able to do the high-value work that’s been avoided because it was too hard.

Read the full post →

ICYMI: Three [Un]Churned Episodes Worth Adding to Your Queue

With a long weekend for some of you, I wanted to resurface some recent UnChurned episodes. They’re less about trends and more about the structural decisions that actually move retention, scale, and leverage.

Ep. 161 — Stop Managing 38% of Your Customers

ft. Yvette Hill Smith (F5)

In this episode, Yvette makes a compelling case for something most CS orgs resist: intentionally not managing a meaningful portion of their customer base. We unpack how over-coverage quietly drains teams, why segmentation often fails in practice, and what it really takes to let go without putting retention at risk. If your team feels stretched no matter how many CSMs you add, this one hits close to home.

Ep. 159 — The #1 Thing Great CCOs Do in Year One

ft. Alexis Hennessy (Heidrick & Struggles)

This conversation is a reality check for anyone stepping into a Chief Customer Officer role. Alexis breaks down what the strongest first-year CCOs actually do (and it’s not launching sweeping transformations). They narrow the field, establish clarity, and resist the pressure to fix everything at once. If you’re inheriting a messy org and feeling the clock ticking, this episode offers a calmer, more disciplined blueprint for earning credibility early.

Ep. 158 — How Community Increases Retention 4x

ft. Erica Kuhl (Gainsight)

Community often gets positioned as a “nice-to-have.” Erica breaks down why that framing is wrong and how community becomes a force multiplier for retention when it’s treated as infrastructure, not a side project. We dig into where community actually shows up in the customer journey and how it changes the economics of CS at scale.

Wrapping Up

The good scales when you design for sequence and ownership. The bad shows up when work is fragmented and avoided. And AI doesn’t change the equation, it just accelerates the outcome. The teams that win won’t be the ones with the flashiest tools. They’ll be the ones willing to redesign the work itself.

See you next Monday 🧠

🌱 Josh

SVP, Strategy & Market Development @ Gainsight

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Writing about what’s possible is one thing. Building it yourself is another. At some point, you have to eat your own words. We at Gainsight wanted to beef up our own AI transformation, not just talk about it.

So I made leading that effort a top priority in my personal OKRs. In early November we assembled a CS working group and started by understanding which frictions we faced the most as a CS org.

We came to the consensus that Instance Reviews (IRs) would be the best place to start. That wasn’t obvious to me at first. I was relatively new to the role and didn’t even know what IRs were yet—but I learned quickly.

The Role of Instance Reviews in Customer Success

Instance Reviews are one of the most important activities a Customer Success Manager (CSM) can do. They’re deep assessments of how customers are using Gainsight. They answer the questions every customer has: Am I getting the most out of this? What should I improve? What best practices am I missing?

They influence Quarterly Business Reviews (QBRs) and ultimately drive renewals, catch risk early, and identify expansion opportunities.

The problem? Few people on the team felt comfortable doing them. Instance Reviews require a level of technical analysis that not every CSM is trained for, and without a clear, repeatable approach, the work felt intimidating rather than empowering.

Roman Dalichow, our VP of Customer Success, put it bluntly: “A whole swath of CSMs shrivel up when they hear ‘instance review’.”

We knew we had to fix this, and AI was an obvious place to look. But in my industry conversations, I’d seen too many AI projects fail to make it past the demo phase. We didn’t want to throw technology at a problem we didn’t fully understand.

So we brought in Method Garage.

Why We Didn’t Do This Alone

We knew what we wanted to solve, but we also knew we didn’t want to spend months rediscovering things others had already learned. This work needed momentum, pattern recognition, and a level of experience we didn’t yet have internally.

That’s where Method Garage came in.

They’re a long-standing partner of Gainsight, and what makes them effective is what I think of as triple fluency. (1) They understand Customer Success because they’ve spent the last decade working with CS teams at enterprises like VMware, DocuSign, and Salesforce. (2) They understand what’s actually possible with AI today because they build real systems that run in production. (3) They’re builders first, focused on writing code rather than producing slide decks.

We brought Method Garage in to run a Blueprint and Prototype Sprint. At HYPER SPEED. One week to map how the work actually happened. Two more weeks to build something real.

Here’s what we learned.

What I Thought I Knew

I came in with a set of assumptions about what applying AI to Instance Reviews would require.

At the time, I believed:

• Instance Reviews are a single end-to-end process we needed to automate

• Nobody had cracked this yet, so we’d be starting from zero

• We’d need a data warehouse or complex integrations to feed the AI

• The payoff would be efficiency: 8-12 hours down to under an hour

I was wrong on almost all of it.

Finding #1: IRs Involve Two Completely Different Activities

We’d been using the term “Instance Review” as if it described a single, end-to-end process. What we quickly learned is that it’s actually two distinct jobs that happen to share a name.

The first layer is technical analysis. This is about understanding what exists in the customer’s Gainsight instance and how it’s configured:

• How is the customer’s Gainsight instance set up?

• Which features are turned on and which are unused?

• How many rules and reports are in place?

• Are scorecards current or stale?

• Are they using Journey Orchestrator only for email, or have they built more sophisticated, multi-channel programs?

The second layer is the value narrative. This is where the technical details are synthesized into something meaningful for a customer conversation:

• Is this configuration actually working?

• Is the customer getting full value from their investment?

• What should they improve or clean up?

• Which best practices are they missing?

• What’s the “so what” that matters to the people on the other side of the table?

These two layers require very different skills. Some CSMs are exceptional at telling the story and connecting insights to outcomes, but feel less confident navigating a complex configuration. Others can quickly read an instance and spot issues, but struggle to translate that analysis into a clear, customer-ready narrative.

We had been trying to automate one process. What we really needed was to support two.

Finding #2: We Had an Elliot

It turned out we didn’t have to look far. One of our Principal CSMs, Elliot Hullverson, who’s been at Gainsight for more than eight years, had already figured out how to make Instance Reviews both fast and effective.

He can complete the technical analysis portion in about 15 minutes. From there, he still spends time building the value narrative and creating the deck, which brings the total to a few hours for him.

For most others, the same work takes ten to twelve hours.

Elliot was proof the problem of preparing Instance Reviews was solvable. We just hadn’t learned from him nor scaled his process.

Our broader lesson was that every team has someone (or some combination of people) who’ve figured out pieces of the problem. Sometimes the gold is buried across three different heads. Method Garage is exceptionally good at finding these people and extracting what’s in their brains. When the expertise is fragmented across multiple people, that’s even harder to do yourself.

Finding #3: The Data Was Already There

I see companies constantly falling into the trap of, “We can’t do AI until we get our data warehouse running. We need to centralize everything. We need to fix our data nightmare first.”

Guess what? Companies have been trying to fix their data problems for decades. If you wait for perfect data infrastructure, you’ll never build anything.

I can say from personal experience: it didn’t matter.

Method Garage found data from various places that nobody was even using. Our CSMs were manually logging into customer instances and reviewing them screen by screen by screen, even though much of that same data was sitting in a 25-megabyte Excel export that our admin tool generates.

We assumed we’d need a data warehouse, complex API integrations, and ample engineering time to build data pipelines.None of that was true. The config export plus usage data from our Gainsight PX integration gave us 80-90% of what we needed.

Before you scope out a data infrastructure project, take inventory. The data you need might already exist in an export you’ve never looked at.

Finding #4: Time Savings Wasn’t the Real Value

Our original pitch centered on efficiency. By applying AI, work that once took 8–12 hours could be completed in under an hour.

But as discovery progressed, it became clear that time wasn’t the real constraint. The deeper issue was that Instance Reviews were cognitively heavy, context-switching-intensive, and emotionally draining. They required deep product knowledge, data interpretation, narrative construction, and executive-ready storytelling—all layered on top of an already overloaded CSM role.

The process was so painful that CSMs weren’t doing them at all.

That reframed the problem entirely. During a working session, I put it this way: “This is no longer about faster Instance Reviews. This is about going from zero to one. It’s the difference between doing them and not doing them.”

The true value came from enabling a task that was effectively non-existent.

The temptation with AI is to focus on efficiency gains, shaving time off tasks that already exist. This framing misses the more important shift.

AI can make an enormous impact by making previously impractical work feasible. When effort drops below the threshold of resistance, teams stop avoiding the work and start building it into their rhythm. At that point, AI is no longer an accelerator. It becomes an enabler of entirely new habits.

The Blueprint Process

The first week of our Blueprint sprint was focused on understanding how Instance Reviews actually happened, not how we assumed they worked.

With Method Garage’s help, we ran eight internal interviews with everyone from individual CSMs to our VP of Customer Success. Instead of asking, “What do you need?”, we asked something more revealing: “Show me how you do this, and explain why.”

That distinction mattered. Watching the work in real time surfaced details that never appear in documentation.

Doing this internally would have taken months. It was Q4, calendars were packed, and context switching would have slowed everything down. Having a dedicated team focused on discovery created momentum we wouldn’t have had otherwise.

Method Garage helped us get past surface-level explanations and into the real decision-making behind the work. Why did you choose that? What were you looking for? How did you know when to move on?

Those answers live in people’s heads, especially your strongest practitioners. They’re rarely written down, and they’re exactly what needs to be captured to automate the work responsibly.

The work done in that week produced the following results:

• They found internal experts like Elliot who had already developed repeatable approaches.

• They mapped the real workflow, not the idealized version.

• They surfaced friction across the work, from technical analysis to turning insights into a clear customer story.

We left with a current-state workflow diagram, a future-state design showing where AI could help, and a clear plan for what to prototype next.

What We’re Planning to Prototype

With the current state mapped, we moved into the future-state design phase.

Here’s what changed from our original thinking: we stopped thinking about “Instance Review Automation” as a single tool. The discovery showed us this is two distinct jobs with different skill requirements.

So we’re designing two separate modules that solve each job (Config Intelligence and Value Narrative), and that can be reused for other CS workflows down the road.

More on this in Part 2 of our story coming soon....

Lessons for Other Leaders

What we learned in that first week applies beyond Gainsight. If you’re thinking about automating a CS workflow (or any knowledge work), here’s what I’d take from this experience.

1. Know Your Real Processes

We’re a Customer Success company, and many of our CS processes are well-documented, consistent, and repeatable, just as you’d expect. But not all of them. Instance Reviews were one of those gaps. They lived largely in tribal knowledge and individual habits, which meant we didn’t understand the process at the level of detail required to automate it.

It’s a common reality you have to confront before you try to apply AI.

2. Find Your Elliot(s) Before You Build

I talk about this quite often. Someone on your team, or more often a small group of people, has already solved pieces of the problem manually. Find them. Learn from them. Bring them into the design process early.

If you automate without them, you’ll automate the wrong thing.

When expertise is scattered across multiple brains, extracting it becomes even harder. That’s where an experienced partner can make a real difference by helping surface what already exists.

3. Ask to See the Work and Why

People are good at describing aspirational processes. Observation reveals how work actually happens.

But observation alone isn’t enough. You also have to ask why. Why did you take that step? How did you make that decision? What were you looking for at that moment?

The reasoning behind those choices is rarely documented, yet it’s exactly what needs to be captured if you want an agent to do more than automate the surface.

One thing that helped us was mapping the entire workflow visually. Four phases. Thirteen steps. Laid out end to end. That diagram became an alignment tool. People who weren’t close to the Instance Review process could finally see how it really worked and how much it varied from person to person.

4. Challenge Your Framing Early

We originally framed this as a time-savings problem. Eight to twelve hours down to under an hour.

What we learned is that it was really a capability problem. The biggest win wasn’t doing the work faster. It was enabling work that often wasn’t happening at all. That shift changes the entire business case.

5. Start With the Data You Have

The instinct to fix your data infrastructure before starting an AI project has derailed more initiatives than bad models ever have.

Take inventory of what already exists. You’ll probably find you can get much further than you expect without waiting for perfection.

6. Blueprint Before You Build

That first week of interviews saved us from building the wrong thing. It also created buy-in from the people who would actually use the solution. Elliot will champion this work because he helped shape it.

Discovery isn’t a delay. It’s de-risking.

What’s Next

With help from Method Garage, we’re building the prototype over the next two weeks.

Yes, I said two weeks. Sounds crazy, I know.

One week to understand a process that’s been running (and frustrating people) for years. Two more weeks to have something working. Our focus is to test the process against real customer data and compare outputs to what Elliot produces manually. We’ll see if the AI can actually find the same insights he finds.

To be clear, a working prototype doesn’t mean the work is finished. It shows the path is viable, reduces risk before a full build, and creates momentum internally.

After the prototype, we’ll make a clear go or no-go decision on building this for real at a production grade. From there, we’ll iterate as we roll it out to more CSMs, knowing it won’t be perfect on day one. Real users always surface real edge cases, and improvement comes from working through them.

If the prototype works, we’ll know the path is viable. We’ll have evidence instead of theory, and a much clearer picture of what it actually takes to deliver Elliot-level capability to every CSM.

Part 2 is coming soon, where I’ll share what happened when we actually built it.

🌱 Josh

SVP, Strategy & Market Development @ Gainsight

Want to compare notes with other Customer Success leaders experimenting with AI and agentic workflows? Drop a comment below or reach out. I’d love to hear what you’re building.

A huge thank you to Method Garage for partnering with us on this work. Their ability to uncover real workflows, surface hidden expertise, and turn messy reality into something buildable made this progress possible. If you’re exploring how to apply AI to complex, human-heavy processes, they’re worth knowing.

👋 Connect with me on LinkedIn

🎧 Listen to more episodes of [Un]Churned

Here’s what’s been on my mind this week.

3 CS Trends That Will Define 2026 & Kristi’s Next Chapter ft. Kristi Faltorusso (This Week’s [Un]Churned 🎙️)]

I had Kristi Faltorusso on Unchurned this week, and honestly, this one was just a lot of fun to record. Every time Kristi comes on, it feels less like an interview and more like catching up with someone who actually gets it. We laughed, we challenged each other, and we ended up covering way more ground than I expected going in.

Kristi came on at an interesting moment in her career. After five years in her role as CCO at ClientSuccess, she decided it was time for something new, even though things were going well. That decision set the tone for a candid discussion about leadership, timing, and choosing change before it’s forced on you.

From there, we dug into the trends she thinks will shape Customer Success in 2026. We spent a lot of time on why CS Ops is becoming one of the most important roles in post sales, especially as AI starts to separate teams that can scale from teams that can’t. We talked about tech stack consolidation, why it’s inevitable, and what that means for CS leaders who’ve been living with tool sprawl for years. And we got into how CS roles are changing faster than job descriptions can keep up.

This episode isn’t a polished theory or hot takes for the sake of it. It’s a real conversation between two people who’ve been in the work and are trying to make sense of what’s coming next. If you want something thoughtful, honest, and genuinely enjoyable to listen to, this one’s worth your time.

Retention Is Invisible Until You Put a Number on It

Kelly McGuire calls out a quiet risk heading into 2026. A CSM can do everything right on retention and still look replaceable if they can’t speak to revenue impact. When churn doesn’t happen, ARR stays flat, and on paper it looks like nothing changed.

This becomes even more important as AI reshapes Customer Success. As automation raises the bar, leaders will prioritize roles that can clearly explain what revenue was protected and what risk was removed. Retention is risk management, but only if you can quantify it.

The takeaway is simple. If retention work stays invisible, it won’t be rewarded. Every save needs a number, and every leader needs a story finance they can repeat.

Read the full post →

The Part of SaaS That Survives AI

Glen McCracken pushes back on the idea that agentic AI means the end of SaaS. The confusion, he argues, comes from treating the interface as the product. Screens were never the core of SaaS. They were simply the most visible layer.

As AI agents take on more execution, the real work moves underneath. Data models, permissions, auditability, contracts, and integrations matter more, not less. Agents don’t remove the need for SaaS foundations. They raise the bar for governance, accountability, and control.

What’s useful here is the reminder of where the real complexity actually lives. As the interface fades, systems of record, trust, and accountability do more of the heavy lifting, even if they’re no longer front and center.

Read the full post →

A Reality Check on CS in 2026

A lot of the conversations about 2026 are starting to converge, and this article from SuccessCOACHING reflects that shift. Customer Success is being pushed to grow up fast, with clearer ownership of outcomes, tighter ties to revenue, and far less tolerance for vague value statements.

AI shows up throughout, but not as a shortcut. The expectation is that automation raises the bar rather than lowering it. Teams with strong fundamentals will scale faster. Teams without them will struggle to explain their role. It’s a solid pulse check on where Customer Success is headed as we move into 2026.

Read the article →

Wrapping Up

Across the episode and both posts, the throughline is clear. As AI changes how work gets done, the teams and leaders who last will be the ones who can explain impact, manage complexity, and make the invisible parts of the business visible.

See you next Monday 🧠

— Josh

👋 Connect with me on LinkedIn
🎧 Listen to more episodes of [Un]Churned

Over the past few weeks, I have spoken with many of you regarding the security advisory involving Gainsight’s Connected App on Salesforce and promised a comprehensive technical explanation. This post explains what we found, what the investigation revealed, and what the Gainsight team built in response.

How the Advisory Began

On November 19, 2025, Salesforce contacted us about unusual activity involving Gainsight-published applications connected to Salesforce. They saw requests hitting their systems that weren’t coming from Gainsight’s applications, infrastructure, or our IP addresses, but were using OAuth tokens that had been issued to the Gainsight-Salesforce connector.

We immediately engaged Mandiant and CrowdStrike, two of the world’s leading cybersecurity firms, to conduct a comprehensive investigation. They had full access to all our environments: production, development, and corporate; all systems, endpoints, backups, and logs. There were no constraints on their scope.

Separately, on November 20, 2025, Mandiant received an email from an individual or group self-identifying as threat actor(s) containing a file with 285 Salesforce OAuth tokens associated with the Gainsight integration.

After completing their exhaustive investigation, they concluded three key things:

1. They could not identify the source of the leaked token set.

2. They found no evidence in Gainsight logs of active threat actor(s).

3. They found no evidence, going back as far as our logs allow (one year), that the leaked tokens originated from Gainsight’s systems.

The investigation established the timeline and nature of the events, but identifying the root cause was more complex due to the timeframes involved, which we’ll explain.

Two Events Across Two Years

What appears as a single advisory actually involves two distinct events separated by at least two years.

Recent event (October — November 2025): Threat actor(s) reused older OAuth tokens to call Salesforce APIs against customer orgs where those tokens remained active.

Historical event (~August 2023): Threat actor(s) originally obtained those tokens from Gainsight or harvested them from external environments or endpoints outside of Gainsight’s control.

To understand what happened, we had to work backward. We started with what we could see clearly, the recent token usage, and traced back to determine where those tokens came from.

Analyzing the Token Usage

On October 22, 2025, the threat actor(s) validated approximately 250 OAuth tokens associated with the Gainsight integration, testing in bulk to determine which credentials were still active. This is standard attacker behavior: you don’t use old credentials at scale without first checking what still works.

Between November 16–19, 2025, the threat actor(s) used the validated tokens to call Salesforce APIs against customer orgs where those tokens remained active. This activity occurred on Salesforce infrastructure, and we did not observe corresponding access to Gainsight applications, systems, or APIs as part of it.

On November 19, 2025, Salesforce detected the suspicious activity and responded by revoking all active tokens associated with the integration and temporarily removing the Gainsight connector from the AppExchange. This was the appropriate response to limit further exposure.

Analyzing the Token Origin

Mandiant’s investigation focused on the harder question: when, where, and how these tokens were originally obtained.

The story here is more complicated. The first step was to determine where the set of tokens existed in the Gainsight environment. We did not find this set in our current systems or in any of our backups.

Mandiant examined the 285 Salesforce tokens in the threat actor(s) dataset and determined they were at least two years old. The most recent token was created in August 2023; the oldest dates back to October 2017, reflecting customer-issued Salesforce OAuth tokens that can persist until customers rotate or revoke them.

One finding stood out. By late November 2025, only about 23% of those tokens still existed in our systems. The rest had been rotated away over time as customers updated credentials. This is a significant finding. It confirmed that the token set was a historical snapshot, ruling out a recent breach. The only window during which this entire set of tokens existed anywhere in our systems, was in and around that historical period, more than two years in the past.

The threat actor(s) behavior reinforces that conclusion. Testing credentials in bulk on October 22 is what you would expect when someone is sitting on an old dataset and trying to figure out what still works.

Where did these tokens come from?

Given the age of the tokens, there are two possible origins:

1. They could have leaked from Gainsight’s own systems at some point in the past.

2. They could have been harvested from external environments or endpoints outside of Gainsight’s control.

Since we could not definitively determine the source, we proceeded on the stricter premise that the tokens originated from our systems.

However, time is an unavoidable constraint. Our logs go back one year, which is consistent with industry standards for security, privacy, and compliance. Within that window, Mandiant found no evidence that would explain how these tokens were obtained from our systems.

Because the tokens are aged between two and eight years, the original exposure event occurred outside our forensic visibility. While we cannot rewind the clock to identify the specific historical vector, we can address why a token from 2017 was still usable today.

At the Root of the Issue: Long-Lived Tokens

While we cannot determine the exact circumstances of the original leak, we can identify the systemic issue that caused this advisory: OAuth tokens were allowed to live too long.

Unlike physical key cards that expire when you check out of a hotel, these tokens could remain valid indefinitely unless explicitly revoked or rotated. A token issued in 2017 could theoretically still work in 2025.

This wasn’t unique to Gainsight. Long-lived OAuth tokens were common practice across the industry, and for understandable reasons. Frequent token expiration creates friction: integrations break, users get logged out, automated workflows fail. The original design prioritized seamless access.

Today, the security model has shifted from “keep tokens safe” to “assume tokens will eventually leak and limit the blast radius when they do.” If tokens can live for years and are leaked at any point in that lifespan, they would remain exploitable far into the future. The window of vulnerability becomes the entire lifetime of the token.

That’s why we focused on solving this systemic issue immediately.

Immediate Remediation: What We Built in Weeks

When the advisory emerged, we immediately reprioritized our engineering efforts to accelerate security hardening and risk reduction across the platform, while continuing to support our customers and ongoing product development.

Access and Credential Security:

• Rotated all system credentials and tokens to eliminate any potential compromise

• Enforced multi-factor authentication (MFA) for all users

• Deleted unused or stale access keys

• Deployed solutions to scan for and secure sensitive credentials

OAuth Token Lifecycle Management: Our CEO Chuck Ganapathi wrote in detail about the four major OAuth enhancements we implemented: automatic token rotation (tokens now refresh every few hours), refresh token protection (each refresh token works only once), trusted IP range restrictions, and Proof Key for Code Exchange (PKCE) for the authentication flow. These protections would now block, or materially limit, the attack pattern we observed.

Beyond OAuth and credential management, we used this moment to strengthen several foundational layers of our security stack:

Data Protection:

• Expanded encryption-in-transit coverage across internal load balancers

• Enabled and enforced encryption for message queues across environments

• Hardened storage policies to prevent overly permissive access

Network Security:

• Implemented geo-restrictions at our Web Application Firewall, API Gateways, and content delivery networks

• Restricted access to cloud resources to only essential personnel and services

Monitoring and Detection:

• Enhanced audit logging to capture more detailed security events

• Improved alerting capabilities to detect and respond to threats faster

In parallel, we broadened our security review beyond the scope of this advisory and made additional hardening improvements across the platform.

From Remediation to Long-Term Hardening

Beyond immediate remediation, we’re executing a comprehensive security roadmap that raises standards across our entire platform. This work focuses on structural improvements that reduce risk across product, infrastructure, and operations.

1. Product Security: Completing CASA Tier 3 assessment, meeting ASVS level 2 standards, embedding security earlier in development, and increasing security gates to proactively detect and address flaws before production.

2. DevSecOps: Working toward SLSA level 3 for software supply chain security, hardening cloud configurations, enhancing secrets management in pipelines, and reducing credential lifespans across the board.

3. Enterprise Security: Reviewing and updating our overall security strategy, increasing training frequency, enhancing privileged access controls, and refining vulnerability management processes.

4. Operations: Strengthening zero trust architecture, increasing real-time monitoring capabilities, enhancing detection systems, and increasing the frequency and rigor of incident response testing.

What This Means Going Forward

Gainsight data is secure. Based on the investigation and available telemetry, we found no evidence of compromise of Gainsight systems or customer data exfiltration from our environments. Salesforce detected and contained the suspicious OAuth token activity on their side.

The systemic issue has been addressed. We have revoked legacy tokens, implemented frequent rotation and additional protections, and the pattern observed here would now be blocked or materially limited.

The platform is materially stronger today. We accelerated security enhancements across our roadmap, and that work was validated by independent third-party experts who investigate compromised systems for a living.

The Response and the Road Ahead

The investigation has concluded, and the priority security hardening work is substantially complete. Salesforce has reconnected our integration, and we are now focused on restoration and ensuring every customer is fully operational.

Some advisories have a tangible root cause, like a phishing email or a misconfigured server. This one did not. It combined a systemic issue around token longevity with a historical event that fell outside our forensic visibility. What we can say definitively is that the systemic issue is fixed, the platform is secure, and it is significantly hardened.

Throughout this process, we have prioritized transparency, sharing indicators of compromise, holding frequent office hours, and publishing findings as quickly as we responsibly could. That commitment continues, and I hope this article provides clarity on both the incident and our response.

– Prem Parameswaran, CTO, Gainsight