The churn signal was hiding in day 5
Rebuilding trial onboarding around an activation milestone instead of a calendar
Context
A marketing automation platform — email and ads — for SMB ecommerce merchants, with a 14-day free trial as the front door. Trial onboarding was a time-based email sequence: day 1 welcome, day 3 feature tour, day 7 case study, day 12 discount nudge. Tidy, automated, and built around our calendar instead of the merchant's behavior. I owned lifecycle for the trial funnel.
The tension
Retention data showed a cliff nobody was talking about. Merchants who hadn't imported their contact list by day 5 churned at roughly three times the rate of those who had. The logic, once you see it, is brutal in its simplicity: this was an email platform, and you can't send email to no one. No list, no first campaign; no first campaign, no proof of value; no proof, no conversion. Yet the onboarding sequence didn't ask for the list until late in the trial — we were giving merchants a feature tour while the one behavior that predicted their survival sat unprompted.
Median time-to-first-campaign sat at 7 days on a 14-day trial. Half the trial was gone before the product proved it could make anyone money. The data behind all this was imperfect — tracking gaps, manual cohort stitching across two tools — so I shipped the analysis with caveats attached instead of waiting a quarter for clean pipelines. Directional was enough to act.
The decision
The option I killed: writing better email. When a sequence underperforms, the instinct is sharper copy and more touches. But the diagnosis wasn't a messaging problem — it was a sequencing problem. So I rebuilt the lifecycle as behavior-driven: every merchant's journey branched on one question — list imported, yes or no. Pre-milestone, every surface pointed at one action. Post-milestone, the journey shifted to getting the first campaign out the door fast.
The second decision was about that first campaign. Asking a merchant to build one from a blank canvas was its own friction wall. So we paired the milestone with a pre-built, high-value campaign template — abandoned cart recovery, copy and design done, one click to send. Import your list, send money-making email five minutes later. That's the value proposition, collapsed into a single session.
The build
Mapped every touchpoint against the milestone. Cut four emails that served our roadmap announcements rather than the merchant's progress. Built triggered flows on behavioral events: an import-abandoned recovery flow, a milestone-celebration message that introduced the template, and a day-4 intervention for accounts showing fit signals but no list. Partnered with product to move the import step earlier in the in-product onboarding and anchor a checklist on it — so the email program and the product surface finally told the same story.
Drag the day the merchant imports their list. Illustrative model, altered per disclaimer — the sequencing logic is the point.
List imported on day 8
The results
Median time-to-first-campaign dropped from 7 days to 3. Activation — merchants importing a list and sending within the trial — rose from 22% to 34%. Trial-to-paid conversion improved 19% relative in the affected cohorts over the following quarter, and the template-first approach performed well enough that it became the default onboarding path.
What didn't move: email open rates. Up barely a point. That was the most clarifying result of the project — the lever was never the channel or the copy. It was putting the right ask in front of the right behavioral moment. A lifecycle program is a sequencing problem before it's a writing problem.
What I'd do differently
Bring product into the room in week one. The in-product checklist drove a meaningful share of the activation lift, and it shipped two months later than it could have because I framed the project as an email rebuild instead of an activation program.