Conversation
Highlights

Cait Keohane thought she knew what she was walking into.

She had spent 13 years at Zendesk, joining as employee 66 when the company was around $100 million in revenue and leaving when it was close to $3 billion. She built the post-sales motion from scratch, helped take the company public, navigated a private equity acquisition, and watched a scrappy startup grow into a 7,000-person organization. She had a playbook.

Then she joined Airtable as Chief Customer Officer and spent her first 90 days getting it wrong.

“I came out of Zendesk and after 13 years, I kind of thought I knew what I was doing, right? Like, I’ve got a playbook for that. I’m coming to a smaller company. You know, I could do this in my sleep. Well, it was a rude awakening.”

The problem was the product. Zendesk was a best-in-class point solution with a defined use case. Airtable is a horizontal platform — anyone from a non-technical founder to an enterprise ops team can build entirely different applications on top of it. That distinction changed everything about what customers needed from the CX org. “I underestimated how much our customers need us for things like change management, governance, real deep support around the non-tangible things.”

She wouldn’t have caught it from behind a desk. She learned it by sitting shoulder to shoulder with customer-facing teams and doing something that sounds obvious but rarely gets prioritized: requesting time with customers with no agenda other than listening.

When Your CS Playbook Is the Problem

In a recent episode of Unicorn Builders: CX, Cait shared how that listening tour reshaped her entire approach to customer success at Airtable — starting with who she hired and what she asked them to do.

The traditional customer success motion is what she calls air traffic control: managing touchpoints, tracking health scores, coordinating renewals. It’s a coordination role. What Airtable’s customers actually needed was something closer to a technical advisor — someone who could translate specific business problems into platform solutions. The two profiles are not the same.

So Cait made a deliberate shift in the CS org’s skill set. “The traditional role of customer success, which is usually a little bit more like this air traffic control…we said, look, we need to up-level the skills and capabilities within this organization to be more technically fluid.”

This wasn’t a cosmetic adjustment. It changed who she hired, how she developed existing team members, and what she measured. It came entirely from customer listening — not internal strategy.

The Metrics That Actually Matter Before the Renewal

Re-profiling the CS team solved one problem. But Cait also had to rethink how she measured success, because the standard metrics were showing her what happened after customers decided to leave — not before.

GRR and net expansion rate are the numbers she ultimately lives by. But she’s direct about their limitation: they’re lagging indicators. By the time those numbers move, the decision is already made. “The challenge is so many customers make a decision about whether or not they want to continue with a platform way before the contract renewal date.”

Her daily scorecard runs on leading signals: activation rate, feature adoption, services attachment, time to first response, time to resolution, and CSAT. When those are green, the lagging indicators follow.

One source she returns to consistently is the one most teams ignore. Support ticket data — volume trends, sentiment patterns, spike frequency after new product releases — is what she calls possibly the least glamorous place to look for insights, and one of the most reliable. “I think so much of the success of your company lives in the support ticket data.” Spikes in tickets after a launch point to QA issues. Persistent negative sentiment points to something breaking in the customer relationship. Both show up in ticket data before they appear anywhere else.

The AI Problem Nobody Is Talking About

The most unexpected challenge Cait has run into at Airtable has nothing to do with her team. It’s happening inside her customers’ organizations.

Everyone wants to adopt AI. The appetite is real. But the pattern she keeps observing is spike-and-drop: customers engage heavily, then usage falls off. The experiment ends without becoming a workflow. “There is a change management hurdle with AI that I haven’t seen elsewhere.”

This creates a direct problem for any software company with AI features. Activation is not adoption. Getting a customer to try something once and getting them to embed it in how their team operates every day are entirely different challenges — and the latter requires change management support that most CS orgs aren’t built to provide.

Cait is working to solve this at Airtable, but she’s honest that it’s not solved. “How do we help them drive durable AI adoption? So not just experiment and then move off, but how do we build this into their day-to-day workflows such that it then becomes just part of how they run the business.”

For founders: if customers are using AI features once and not returning, the problem probably isn’t the feature. It’s the absence of a structured adoption motion around it.

Build in Real Time or Fall Behind

The throughline across everything Cait does is deceptively simple: stay close to customers and adjust. The Zendesk playbook didn’t transfer to Airtable. The playbook she’s building now may not transfer to the next company. What transfers is the discipline of listening before assuming.

For founders hiring their first CX leader, she has one clear criterion. Find someone who will build in real time, not execute from memory. “AI is the great leveler. There are no playbooks anymore. We’re all building them real time.”