The short version

People rarely leave a note explaining why they’ve stopped using your product. Fortunately, they do leave clues. Look at where they drop off, talk to them about what happened and check whether they ever got the result they came for. You might need fewer questions, clearer guidance or a better fit with what they actually need. Pick the biggest obstacle, make a change and see whether it helps.

The (slightly) longer version

You've worked hard to get people interested in your product. They're visiting the website, signing up and then... not much. Some disappear halfway through getting started. Others try it once and never return.

There are all sorts of reasons for this, and they depend on the product and the customer. So before jumping to conclusions, it's time to put your detective hat on and look for some clues. The first thing I'd want to know is whether those people ever experienced the value they signed up for.

Get people to the "AHA!" moment

I like to think about onboarding in terms of the "AHA!" moment. Imagine someone using your product for the first time and thinking, "Aha! I get it. That's why this exists!"

You understand your product better than anyone. You've spent months, perhaps years, thinking about it. Someone who's just arrived knows next to nothing, even if your website explains it beautifully. They're also busy, with plenty of other things competing for their attention.

I find it useful to think about how likely someone is to carry on using your product as a little calculation:

Willingness to continue = (Perceived value × Confidence) ÷ Effort

How useful does this look? How confident am I that it'll actually do what I need? And how much work is it going to take to find out?

It's not at all scientific, just a useful way to think about the trade-offs people are making.

Your marketing starts building that sense of value and confidence. Once someone arrives in the product, the experience needs to back it up. If they've misunderstood what you offer, that might be a marketing or targeting problem rather than an onboarding one.

Show the value before asking for too much

Some of us have strange hobbies. Apparently one of mine is asking startup founders on Reddit to share their products so I can have a look around and give them feedback.

One thing I've noticed is how much information some products ask for before showing me anything useful. I'll sign up, then find myself working through page after page of questions about me, my business and how I plan to use the product. Hang on, I haven't even seen what this thing does yet!

Some of those questions might be necessary, but asking them all upfront creates a lot of effort before I've experienced any value. Could some wait until later?

Another common issue is dropping someone straight into an empty dashboard and expecting them to explore. Learning a new system takes effort, even if it's obvious to the people who built it. Help them do something useful instead of leaving them to work out where to begin.

Duolingo is a nice example. Its onboarding gets people trying an actual language lesson early, rather than spending ages explaining what learning with the app will be like.

Could you do something similar? Maybe let visitors try an interactive demo on your website, using sample data, before they even register.

Then look at everything you're asking people to do before they get that first result. Do you need all those questions? Could you look up the information for them? Could you wait until later to ask?

Removing a step altogether is often better than making a long form prettier. The point isn't to remove every bit of effort, but to stop spending your user's effort on things that don't yet help them get value.

Find out what's happening in your own product

Rather than guessing which screens need changing, I'd investigate it in this order.

1. Work out what getting value actually looks like

First, identify the action that tells you someone may have experienced what makes your product useful.

A few examples:

• Marketplace: Completing their first successful transaction.

• Activity tracker: Logging enough activity to start seeing useful insights when they review their stats.

• Streaming service: Watching their first episode.

• To-do list: Completing a task they've been keeping track of.

Be careful not to confuse an intermediate step with the actual value. With a to-do list, for example, you might be tempted to measure how many tasks people add. But adding tasks isn't really the outcome they want. They want to get things done, keep track of their responsibilities and free up some mental space. Completing a task is likely a better indicator.

Even then, real life is messy. Someone might watch their first episode on Netflix, but perhaps they stopped paying attention after five minutes and autoplay kept running. A marketplace transaction might be completed but later disputed or refunded. Someone might mark a task as complete simply because they wanted it off their list.

From an analytics perspective, these could all look like successes, even though the customer might not have received the value you expected.

You can improve your metrics by getting closer to the outcome you actually care about. For example, instead of measuring all completed marketplace transactions, you might track transactions that were not disputed or refunded within seven days, while keeping an eye on refund rates separately.

It still won't be perfect, but it's a better indication that the customer actually got what they wanted. Start with the best measure you can reasonably track, then improve it as you learn more.

2. Find out where people are dropping out

Once you've identified that valuable action, look at how many people actually reach it.

I'd start with a few simple questions:

• How many people sign up?

• How many complete the steps needed to reach that first valuable action?

• Where do the others stop?

• How many people come back and experience value again?

A simple funnel in PostHog can show you how many users make it through each step.

If lots of people are abandoning the same screen, that's a good place to investigate. Perhaps the effort is higher than expected, or something is undermining their confidence that they'll get a useful result.

You can also use session recordings to watch what people actually do in your product. Maybe they're repeatedly clicking something that isn't clickable, going backwards and forwards between screens or spending ages staring at a particular form.

That gives you a much better idea of what's happening, but there's still a problem. You can see what they're doing, but you don't know what they're thinking.

3. Talk to the people behind the numbers

I'd speak to a few people who completed the journey and, just as importantly, a few who disappeared along the way.

The trick is to reconstruct what actually happened, rather than asking whether they liked your product.

Ask them to walk you through the last time they tried to use it, blow by blow. What prompted them to visit? What were they hoping to achieve? What did they do first? What happened next?

As they describe each step, dig into the little decisions they made. What did they expect to happen? What were they thinking when they saw that screen? Why did they choose one option over another? Was there anything they particularly liked, or something that made them hesitate?

You're trying to understand their motivations, expectations, frustrations and the moments that helped them carry on or made them want to stop.

Try not to lead them towards the answer you expect. Asking "Was that form too long?" might get you a polite yes, but it doesn't tell you much. Discovering that someone stopped because they didn't have their account details to hand is far more useful.

If they can show you what happened as they talk through it, even better.

By now you've got three things: a metric showing whether people might be getting value, data showing where they're struggling, and conversations helping you understand why. That's a much better starting point for making improvements than simply guessing.

Use what you've learnt to make things better

Now bring what you've discovered back to the three parts of our formula: value, confidence and effort.

If people want the outcome but abandon a lengthy setup, you probably have an effort problem. Remove unnecessary questions, delay tasks until they're needed or do some of the work for them.

If people aren't convinced your product will work, focus on confidence. Could you demonstrate the result using sample data? Help them achieve something useful sooner? Explain what will happen next so they're not wondering whether they're wasting their time?

If people don't see why your product is useful, investigate the perceived value. Perhaps you haven't demonstrated the benefits clearly enough, or your marketing is attracting people whose problems you don't actually solve.

But what if people are reaching that first valuable action and still not coming back?

This could be a much bigger problem than a few unnecessary signup screens. First, are these actually the customers you're trying to attract? If not, you might need to improve your marketing or targeting.

If they are your target customers, I'd want to understand exactly what they were hoping to achieve and how that compares with what they actually got. Perhaps your product doesn't quite fit their situation. Maybe they need an integration with a system they already use, an extra step in the process or a feature you haven't built yet.

Or perhaps they got exactly what they needed and simply don't have a reason to return for another month. Not every product needs to be used every day.

This is where those customer conversations become particularly valuable. You want to understand the gap between what users expected and what they experienced, then decide whether it's worth closing that gap and how.

Whatever you discover, resist the temptation to immediately start building more features. Sometimes that's absolutely the right answer, but you should understand the problem first.

Pick the biggest problem you've uncovered, make one sensible improvement, and measure whether more users reach the outcome afterwards. Then repeat.

You might find that the best improvement you can make is helping people experience the value that's already there.