One note per trial user, naming where they actually stopped

What is the exact prompt?

Daily, mid-morning:

• "From PostHog, list every trial account created 2–10 days ago and classify by behavior: ACTIVATED (completed [your activation event, e.g. created a project and invited a teammate]), STUCK (signed up, returned at least once, never completed activation; note the last screen they touched), BOUNCED (one session under five minutes, never returned), BLOCKED (hit an error event or a permissions wall; name the event)."
• "Drafts: for each STUCK user, one email referencing where they actually stopped ('you set up the workspace but haven't connected a data source; that step trips people up') with the one link that gets them past it. For each BLOCKED user, an apology-plus-fix note naming the error. For ACTIVATED users on a team domain, a short note offering a setup call, with no feature pitch. For BOUNCED, nothing; add them to a count, not a campaign."
• "Rules: one email per user per trial, ever. Check the sent log first. Plain text, from me, two sentences to five. No 'just checking in', no urgency theater about trial expiry. Leave all drafts for my review, STUCK first."

BOUNCED gets nothing, on purpose

The ninety-second visitor already gave you their answer, and emailing them trains your domain into spam filters. Not emailing them is what keeps the rest of the play deliverable.

The count still matters: a rising bounce rate is a signup-flow problem, and it routes to the product team through your churn analysis. BLOCKED is the inverse, the smallest and highest-intent segment, visible only because the analytics read sees error events.

Why one reviewed note instead of a smarter sequence?

Reading fifteen drafts takes ten minutes and catches the analytics misread, as when a user reads STUCK because their teammate activated the account. It also keeps you aware of what trials are doing, which is product feedback the sequence era threw away.

The one-email-ever cap is the architecture, and the plain-text note from a real person is what gets replies. Every STUCK reply is a usability report from someone mid-struggle, and threading those into your feedback summaries closes the loop.

Measure replies first, conversion last

Reply rate, then activation-after-outreach, then trial-to-paid, in that order, because the chain fails at the first link. A STUCK note that gets no replies is misreading where users stall, so fix the activation definition before judging the funnel.

Expect the segments to teach you things. The common discovery is that the activation event was wrong: users who "activated" by the metric still bounced, and users who never triggered it converted anyway. That finding tends to repay the play before any email does, the same instinct as qualifying inbound leads by behavior instead of form fields.

Frequently asked questions