Why People Start Your Trial and Never Come Back

Why do people start a trial and never come back?

Because nothing happened in the first session. Most trials are lost before any email sequence begins, at the moment someone signs up, looks at an empty interface, and cannot see the path from here to the thing they came for. Everything after that is an attempt to recover a first impression.

This is the part of the funnel where marketing and product stop being separable. You can fix the landing page, the pricing page and the email cadence, and still lose people, because the thing losing them is the distance between signing up and understanding.

Here is what actually moves trial to paid, in the order I would work on it, and the lever that gets far more attention than it deserves.

What is the one thing that predicts conversion?

Whether the user reached a moment where the product visibly did its job. Every product has one, and it is specific: a report generated, a message sent, a team member invited, a site published. Users who reach it convert at a completely different rate from users who do not, regardless of anything else you do.

The reason this matters more than any other number is that it is actionable. A conversion rate tells you how you did. An activation rate tells you where people stopped, which is a thing you can go and fix on Monday.

Most teams have not written down what their moment actually is. If you ask three people at the company and get three answers, that is the real finding, and defining it properly will improve more than any tactic you could deploy afterwards.

Which levers actually move trial to paid?

Four, in order of impact. Reducing the work between signup and first value. Putting real data in front of the user immediately. Removing the setup steps that can wait. And following up based on what the person actually did rather than on how many days have passed.

The first one is almost always the biggest and almost always the least popular, because it means cutting steps that somebody added for a reason. Every field on a signup form, every configuration screen, every invitation prompt before the product has proved anything is a place people leave.

The second is underused. A blank product is a request for work, and nobody signed up hoping to do work. Sample data, a template, or an import that runs in the background all change the first screen from an empty room into something a person can react to.

Should the trial be time-limited or usage-limited?

Time-limited if your value is obvious quickly, usage-limited if it takes real work to set up. A fourteen day clock on a product that needs a week of data collection is a clock that runs out before the product has had a chance to be useful.

The usage-limited version has a second advantage, which is that it survives holidays, busy weeks and the person who signed up going on leave. A free tier with real limits keeps working while attention is elsewhere, and attention is the resource most trials lose to.

Linear's pricing page is a clear example of the usage-limited shape. On the day I looked, its Free plan was listed at zero dollars with unlimited members, two teams and 250 issues, which lets a team genuinely use the product and hit a boundary that means something rather than waking up to an expired account.

Should you require a credit card up front?

It depends on which problem you would rather have. Requiring a card reduces the number of trials and raises the quality of the ones you get. Not requiring one does the opposite. Neither is right in general, and anyone who tells you otherwise is describing their own product.

The question I would ask is what your team does with a trial user. If a human follows up on every signup, a card requirement protects that person's time and is probably worth it. If the trial is entirely self-serve, the card is friction with no corresponding saving.

What I would avoid is the worst of both, which is requiring a card and then also gating the useful parts of the product behind a sales conversation. That asks for commitment twice before delivering anything once, and it is the arrangement most likely to produce a trial nobody finishes.

What should happen in the first ten minutes?

One visible result, produced with as little input as possible. Not a tour, not a checklist, and not a video. A thing the product made, that the user can look at and judge. Everything else in onboarding is support for getting to that moment faster.

Product tours are the substitute most teams reach for, and they mostly measure well and convert badly. A tour teaches the interface. What a trial user needs is evidence, and evidence means output, not orientation.

The test I would apply is whether a person could close the tab after ten minutes and still describe to a colleague what your product did for them. If the honest answer is that they could only describe what it looked like, the first session is not doing its job.

What should the follow-up actually say?

It should respond to behaviour. Someone who imported data and stopped needs a different message from someone who signed up and never returned, and both need something different from someone who reached the value moment and is now evaluating price.

The default is a fixed sequence on a timer, which treats all three the same and is mostly ignored by all three. Behaviour-triggered messages are more work to build and they are the difference between an email that arrives at a useful moment and one that arrives on day three because it always arrives on day three.

The content matters less than the timing, but it does matter. The most effective trial emails I have seen name the specific thing the person has not done yet and make it easy to do from the email. I went through the structure of that sequence in the onboarding email sequence for B2B trial users. The same behavioural reading tells you who deserves a human message: someone who did a lot on day one and nothing since is stalled, even with ten days left, while someone making slow weekly progress is fine.

Where does pricing fit into this?

Later than people think, and it still has to be visible. A user who has reached the value moment will look at the pricing page next, and if that page cannot be understood in a minute, the momentum you built in the product dissipates at the last step.

The specific failure is a pricing page where the tier a trial user would naturally land on is the one described least clearly. If the self-serve tiers have numbers and the relevant one says contact sales, you have introduced a conversation into a flow that was working without one.

I argued the broader version of that in whether to publish your enterprise price or say contact sales. For trial conversion specifically, the rule is simpler: the tier your trial converts into must have a number on it.

What is the most common mistake?

Optimising the acquisition of trials rather than the completion of them. It is easier to get more signups than to fix a first session, so teams do that, and the extra signups fail at the same rate as the existing ones. The funnel gets wider and nothing improves.

The second most common is treating trial conversion as a marketing problem alone. The levers that matter most are in the product, and a marketing team working without product involvement can only address the edges of it.

The third is related, and it is choosing the trial model before understanding the buyer. Whether people should self-serve at all is a prior decision, and it is worth revisiting rather than inheriting. I set out how I would make that call in self-serve trial versus demo request for B2B software.

What should you do next?

Write down the single moment where your product visibly does its job, then find out what share of trial signups reach it. If you cannot measure that today, instrumenting it is the highest value thing you could build this month.

Then sign up for your own trial with a fresh account and a timer running. Ten minutes, no shortcuts, no admin access. Whatever you are still waiting for at the end of that is what your users are leaving over.

If you have trials starting and not converting and you are not sure which end of the problem to work on, tell me what your product does and what happens in the first session. I will tell you where I would look first. Let's chat.