In June, Toy Story 5 hit theaters. Early on, a sleek new tablet named LilyPad arrives in the kids' lives, and Buzz and Woody suddenly have to work out how to stay relevant in a world moving on without them.
Watching this challenge play out, I felt like I was living it. That's exactly the moment customer success is in with AI right now.
I oversee customer experience for our AI and emerging products at Pendo. I've helped scale post-sale organizations through hypergrowth and turn around broken ones. I'm often the person handed a big problem with no plan and asked to fix it.
I'm also championing AI at Pendo, both in the products we sell and in how we work every day. So I say this as a fan: AI isn't going to fix a broken customer success organization.
The pilot phase is over
We've had roughly a year to experiment with AI. We ran pilots, built strategy decks, and demoed exciting ideas to our leadership teams. That time was useful, and it's up.
Boards and senior executives are no longer impressed by what we're planning to do with AI. They want to see it in the bottom line. They want NRR to move.
That means making three shifts:
- From demos to real deployments
- From a pilot phase to a new operating model
- From an AI strategy deck to movement on the NRR line
If we stay in pilot mode, we'll end up right back where we started.
Most CS organizations I talk to have been racing to add AI to their existing processes. When those processes sit on a shaky foundation, AI makes the problems happen faster. CSMs get more noise, more signals, and more calls to action, with the same outcomes as before.
We've been here before. When we first rolled out customer success platforms, we expected them to change everything. Then we spent months buried in alerts we didn't know how to act on, because we hadn't done the foundational work first.
AI is the same story. Bad data in, bad data out.
What a broken CS organization looks like
We all know the retention math. Acquiring a new customer costs significantly more than keeping an existing one. If we get this wrong, AI won't help with the thing that matters most: retaining and expanding our customers.
The most persistent symptom of a broken CS organization is that we're still reactive. Even after all these years, most of us (Pendo included) are reactive in at least some parts of what we do.
The reasons tend to come back to two familiar problems:
- There's no shared language. Different teams track different things, define health differently, and report on disconnected metrics. Leadership focuses on one number while the CS team optimizes for something else entirely.
- The metrics are wrong, or at least inconsistent. You might be working hard to reduce days to onboarding while your executives measure something completely different.
When there's no shared truth about what good looks like, there's nothing solid to build on. Add AI to that, and you get faster chaos.
Build the foundation first
Before you go deeper on AI, you need to get three things right.
1. A shared health language
Go back to basics. Sit down with your cross-functional leaders and run a litmus test: do you all agree on the top-line metrics your organizations are working toward?
Each team can have its own leading indicators and daily activities. At the top, though, you need shared truth and one definition of what good looks like. Without it, AI will just surface conflicting signals faster.
