Trial to paid onboarding emails triggered by behaviour

Aug 22
3 min read

Updated: 4 days ago

Introduction

Most trial email sequences are a calendar: day one welcome, day three tips, day seven case study, day thirteen "your trial ends tomorrow".

Everyone on the list gets the same messages regardless of what they did, which means the person who never logged in and the person using it daily receive identical advice. Behaviour-triggered sequences fix that, and they convert substantially better.

1. Trial to paid onboarding emails should trigger on actions, not dates

The unit of a good sequence is a condition, not a day.

*Has not completed the activation action after 24 hours.* *Completed it but has not returned in three days.* *Using it heavily and approaching a limit.* Each of those is a different person who needs a different message, and the calendar cannot tell them apart.

Start by identifying your activation action — the first real piece of work that predicts retention. Everything branches off whether that has happened.

2. Three branches cover most of it

Not activated. The largest group and the most valuable to fix. Everything should aim at getting them to that first action: the single next step, an offer to set it up for them, or a two-minute video of exactly that task.

Activated but idle. They saw it work and drifted. Give them a reason to return tied to what they already did — the report they generated, the project they started.

Activated and active. These will likely convert. Do not sell hard; remove obstacles. Answer the billing question, mention the annual option, offer a conversation if they are a team.

3. The first message should do one thing

Not welcome them. Get them to the activation action.

A single link to the one next step, in a message short enough to read on a phone in ten seconds. No feature tour, no list of everything the product does, no video library.

Feature tours in message one are the most common mistake, because they feel generous and they increase the number of decisions a new user has to make.

4. Write from a person, and let them reply

A no-reply address wastes the highest-signal channel you have during a trial.

Replies to onboarding emails tell you exactly where people get stuck, in their own words, from users with the problem in front of them right now. That is better than any survey and it arrives free.

Use a real name and a real address, and read them.

5. Handle the trial ending honestly

The end-of-trial message converts, and it should be plain rather than urgent.

Say when it ends, what happens to their data, and what it costs. If someone has been active and simply run out of time, offering an extension is often better business than losing them — you keep a user who has already demonstrated they get value.

Manufactured urgency reads badly to people who are genuinely still deciding, and this audience is paying attention.

6. Stop sending to people who converted

Obvious and frequently broken. A customer receiving "your trial ends tomorrow" after paying undermines confidence in everything else you send.

Every branch needs an exit condition on conversion. Test it by converting a test account and confirming the sequence stops.

7. Ask the ones who lapse, once

A single question to a trial that expired without converting — what stopped you — produces better information than most research, because they have nothing to gain by being diplomatic.

Sort the answers: never got it working, got it working but did not need it, too expensive for the use I had, missing something specific. Only the last is a product gap, and the first is usually the largest group.

That sorting is what tells you whether to fix onboarding or build features.

8. Measure by cohort, not by month

Judge the sequence on trial-to-paid conversion for cohorts of trials that started in the same period, not on a calendar month that mixes finished and running trials.

Compare cohorts before and after a change to the sequence. Also watch the proportion reaching activation — if that moves, conversion follows, and it moves sooner so you learn faster.

Conclusion

Trigger on behaviour rather than day number, and branch on whether the user reached the activation action: not activated, activated but idle, or actively using it.

Make the first message do one thing, send from a real person who reads replies, handle the trial ending plainly and consider extensions for active users, exit the sequence on conversion, ask lapsed trials one question, and measure by cohort.