The first 90 days. Make the check-ins mandatory.

The check-in trap

Our CSM contact ran on a 30/60/90-day rhythm, and the touches were mostly light — a "hey, everything OK?" sort of email. I'm not sure that's the right approach. Honestly, I'm now sure it isn't. This is the most critical part of a SaaS relationship, and it deserves more than a checkbox.

The trap has two sides. On your side, that email is too easy to send — the CSM checks the box, did we do a check-in?, and the dashboard says yes while nothing actually happened. On the customer's side, they're busy, and a check-in that feels optional gets treated as optional. Nobody responds to "do you have any questions?" with the truth, which is usually "we haven't really started."

The positioning is the whole trick. A meeting that's part of the vendor's stated process gets attended; a courtesy call gets deferred. And those meetings shouldn't be status updates — they're where you introduce and re-introduce the sticky features, the ones that wire the product into how the team works: the weekly pipeline review, custom fields, custom list views on the deals page. A new customer doesn't absorb your product in one training. The 30/60/90s are your scheduled chances to deepen the install before the habit window closes.

The 60-inch Samsung in the conference room

Here's the move we got right, and it's the heart of this whole piece. We didn't just tell customers "use the product, make it a daily habit." We got the champion and the sales manager to run their weekly sales pipeline review from the product — up on the 60-inch Samsung sitting in their conference room.

That's the entire mechanism. Once the weekly meeting runs off the product, every rep has a social reason to keep their deals current that no onboarding email could ever manufacture. You're not asking individuals to build a habit through willpower — you're attaching the product to a meeting that already happens, with their manager and their peers in the room. The accountability is ambient. Adoption stops being a favor the team does for the software and becomes the price of walking into Monday's meeting prepared.

It won't be lost on regular readers of this series that the customers who stuck were the ones running a weekly cadence meeting off their numbers. The lesson generalizes well beyond CRMs: find the recurring meeting your product can power, and make powering it the goal of the first 90 days.

Reading fragile vs. durable

So how did you know it was working? You could see it in the data. What does the login and usage trend look like — building or fading? What's the percentage adoption across the team — is it the whole sales floor, or one diligent admin? And the true indicator: is the team marking deals won? That one matters more than any login count, because it means they're using the product to actually manage their sales pipeline — running their business in it, not complying with a rollout. A team that closes deals in your product has moved their work into it. That's what durable looks like.

New accounts need their own pile

Second honest correction: our new accounts flowed into the same daily usage email as every established customer, and they should not have. A mature account that goes quiet for a week is a yellow flag. A six-week-old account that goes quiet is a different and more urgent thing — the habit never formed, and the clock is running.

What we should have built — and what I'd tell any founder to build — is a separate view for the new-account cohort: health, activation, and feature adoption tracked by cohort, with tighter expectations than the established base. The health score machinery is the same; the thresholds and the urgency are not. Manage the new accounts in a different pile than your already-onboarded customers, because the question you're asking about them is different: not "is this account fading?" but "has this account ever truly arrived?"

When a new account wobbles

And when an account that onboarded cleanly starts wobbling in week six anyway? The intervention looked the same as it does at any age: diagnose, then provide solutions. The reasons were the same ones too — bad data, a mis-configured product, new people who needed training. The full diagnosis-first approach is the save play, and it applies in the first 90 days the way it applies in year three. The only difference is the stakes: catch it in the window and you're finishing the onboarding; catch it after and you're attempting a rescue.