why everyone skips your product tour

you've skipped one yourself

you sign up for a new tool because you want to do one specific thing. the screen dims and a bubble appears: welcome, let us show you around, step 1 of 9. you click skip without reading it. almost everyone does.

twenty minutes later you need the export button, and you can't find it. the tour would have shown you. but the tour is gone, and there's no obvious way to get it back, so you search the help center, or email support, or give up.

notice what went wrong there. the tour had the right information. it showed up at the wrong moment, to a person who wasn't ready to hear it, and then disappeared at the exact moment they needed it. that's most of what's wrong with product tours, and the data backs it up more clearly than you might expect.

what the completion data says

Chameleon sells product tour software, so you'd expect it to be optimistic about tours. to its credit, it publishes benchmark data from its customers that is refreshingly blunt about what fails.

two things stand out. the first is that who starts the tour roughly doubles completion. when the person chose to begin, they usually finish. when a timer chose for them, most don't. the second is that length is brutal. each extra step is one more place to lose someone, and by the seventh step almost everyone is gone.

Userpilot's benchmarks point the same way from a different angle. its average onboarding checklist completion is about 19%, with a median around 10% (source). front-loaded onboarding, however it's packaged, mostly doesn't get finished.

why the automatic tour fails

once you put the numbers next to what's happening in the user's head, they stop being surprising.

on first login people have the most motivation and the least context. they came to get something done, and a tour asks them to stop and learn the layout first. that's a bad trade from where they're sitting. the tour also shows everyone the same path, so the person who came to export a report sits through a stop about integrations they'll never use.

then there's memory. a skipped tour is effectively deleted. very few people go looking for a restart link, partly because they don't know one exists and partly because they don't know the tour covered what they now need. and while it runs, it's in the way. a dimmed screen with a modal on top is physically blocking the thing the person came to do, so dismissing it is the reasonable response.

put together, the automatic tour is a small lecture delivered at the start of a meeting nobody asked for. the content can be excellent and it will still mostly be ignored.

what works better

if the tours that get finished are the ones people start themselves, and the ones that get finished are short, then the conclusion almost writes itself. help should arrive when the person asks for it, and it should cover only the question they asked.

you can move toward that in steps, depending on how much you want to change:

1. add a launcher, a help menu that lists short tours so users pick the one they need. completion goes up because the user chose, though they still have to find the right entry in a list.
2. trigger tours on intent instead of time, so a tour starts when someone opens a feature for the first time, not five seconds after login.
3. split long tours into small ones, one per task, each three or four steps.
4. let people just ask. skip the menu entirely, so a user types or says how do i export this, and the right walkthrough starts on the spot.

the last step is the one we built heytappr around. the user asks in their own words, heytappr matches the question to a guide your team approved, and then walks them through it on the live page, saying each step out loud and spotlighting the real button, moving on when they click. every guide it runs is started by the user and limited to their question, which happens to be exactly the kind of tour the benchmarks say people finish.

so should you delete your product tour?

probably not all of it. a very short welcome, one or two steps that say where the main thing lives and where to ask if you get stuck, sets context without taking over the session. that kind of orientation is cheap and mostly harmless.

what we'd stop doing is the long tour that fires on first login, the tour on a timer, and the tour that walks people through features they haven't shown any interest in. we'd also stop counting "tour viewed" as a sign onboarding worked, since the data suggests most views end in a skip.

in their place we'd put task-sized guides for the questions support hears most, a way to ask for help from any screen, and a habit of reading the list of questions nobody has a guide for yet. that list is more useful than it sounds. it's users telling you, in their own words, what to explain next.

how to test this on your own product

you don't have to rip anything out to find out whether this applies to you. take the most common "how do i" question your support team gets (we wrote about finding those in the playbook for how-do-i tickets). build a three-step guide for it that users start themselves, and leave your existing tour running.

after a month, compare two things: how often each gets finished, and how many tickets that question produced before and after. if your numbers look anything like the benchmarks above, the result will make the decision for you.