The Journey: #86 That's Not My Job Anymore

A customer came to me a few weeks ago wanting to "figure out AI" for their CS team. There was no clear mandate, no direction, everyone doing something different with chat tools and calling it a strategy. Figure out AI. Cool, great, love that for me.

So we didn't start with tools. We started with a question nobody asks in these exercises, because it's genuinely uncomfortable to answer honestly when you're sitting in a room with your manager.

What do you complain about? What do you avoid? What do you know you should be doing and just don't, because there's never enough time in the day?

It took a few days just to build the list, and not because the team was slow. It's because nobody had ever asked them to actually catalog the friction before, so they had to stop and think about it in a way that felt almost foreign. Once they did, we had 22 ideas sitting in front of us. Not 22 use cases for a tool, we had 22 real gaps in how they ran their customer journey, places where AI could make a meaningful impact if someone actually did the work to figure out how.

Then came the part everyone skips.

For every idea on that list, we mapped ten things: the business objective, the expected outcome, who owned it, what systems it touched, what data it needed, the actual workflow, the next best action, how it would be tracked, the baseline, and the KPIs that would tell us if it worked. Every idea had to earn its spot on the roadmap by answering all ten, and if you couldn't fill them in, it wasn't ready. It was just a nice idea with good intentions and a lot of wishful thinking.

Then we ranked them, assigned owners, built timelines, and put together an internal comms plan and a resource ask. Half of these ideas needed hands from outside CS, and nothing kills a good idea faster than nobody funding it and everyone assuming someone else is handling it. They walked into that exec conversation not asking for permission. They walked in with a plan that already had momentum behind it, and that changes the entire dynamic of the room.

That's what separates a real AI roadmap from a doc that lives in the drive to die.

Here's where I see the CS community getting stuck, and honestly it's not a knowledge problem. It's a permission problem people give themselves.

Everyone's adopted the basics, a chatbot for drafting, a summary tool for notes, and then they stop, like they finished building their AI house, and forgot that there was a second floor.

The teams becoming indispensable right now are treating AI as infrastructure, not a feature. They're catching churn signals weeks before you’ve thought to even ask the customer and they’ve had to say it out loud. They're building health scores from actual behavior instead of a CSM's gut feeling updated on a Friday afternoon when they're already half checked out. They're walking into business reviews having spent thirty minutes prepping instead of four hours and a small breakdown the night before.

That's not doing less CS. That's doing better CS at a scale that genuinely wasn't possible before, and the gap between the teams doing it and the teams still dabbling is only going to get wider.

So steal my framework.

Before you greenlight any AI use case for your team, run it through these ten fields. Can't fill in all ten? It's not ready; It's just a nice idea with good intentions.

• Business Objective: What business goal does this serve?

• Business Outcome: What specifically changes if it works?

• Business Owner: Who is accountable for this, not just in CS?

• Systems Integrated: What tools does this touch or require?

• Data Leveraged: What data does it need, and do you have it?

• Workflow Details: What does the actual day-to-day motion look like?

• Next Best Action: What happens after the AI surfaces something?

• Visibility and Management: How does leadership see if it's working?

• Baseline: What does the current state look like before AI?

• KPIs: How do you know it's working?

Run your team's "what do we complain about" list through this before you buy another tool.

If you don't have time for a full audit right now, here's where I'd start:

• Churn prediction: flag accounts heading towards churn

• Dynamic health scoring: build the score from behavior, not internal assumptions

• QBR prep: turn hours of deck-building into thirty minutes with a solid talk track

• Call summarization: stop losing verbal commitments to a bad CRM note

• Renewal forecasting: have a real answer when your manager asks about the book

• Sentiment analysis: read the whole account, not just what you feel coming off your last call with you POC.

Pick one. Actually use it for a month before you add a second.

The takeaway: AI isn't a single use case, it’s an operating model.

If your team's entire AI strategy is "someone uses ChatGPT to draft emails," that's not a strategy. The work is building the list, structuring the plan, and assigning the ownership.

The tool is the easy part. There are plenty as we all know.

Closing with Kristi

The CSMs and leaders thriving right now aren't the ones with the flashiest AI stack. They're the ones who did the unglamorous work of mapping their own friction first. If your team needs help running this exercise, that's exactly the kind of work I do in advisory engagements. In the meantime, go build your list and let me know how many you come up with.

See you next Tuesday,

Customer Success. Revenue Follows.