A tour fires on your calendar. A user gets stuck on theirs. Those have never been the same clock, because the tour has to choose when to run before it knows what is about to go wrong — and it cannot watch to find out.
The tour ran on the first of them. Every day after that, the product had no idea what this person had skipped, tried, or given up on.
They are importing data and creating logins. They are not learning the product, they are setting it up, and they know they can come back to this. They cannot.
For the first time, properly, with real data in it. This is the day the walkthrough would have been worth something, and it is eleven days too late.
Step three of seven, dismissed in four seconds six weeks ago. There is nothing in the product that knows they skipped it, so there is nothing that offers it again.
No ticket, no complaint, no completion-rate change. The account is quietly using a third of what it pays for, and the first anybody hears about it is the renewal call.
The tour did its job on day one and was measured as a success. Three of the four days that actually decided this account happened after anything in the product was still looking.
It is a real number and it is measuring a real thing. The real thing is just not whether they can now do the job.
Seven modals dismissed in forty seconds scores identically to seven modals read. One of those people can now do the thing.
Almost everybody skipping a tour intends to come back to it. The product offers them no way to, and no longer remembers they wanted to.
Completion is highest when nobody needs help yet, and there is no equivalent number at all for the week they actually got stuck.
These are properties of being a script rather than shortcomings of any particular product. A smarter trigger changes when it plays, not what it is.
A tour is a script that plays. It has no idea the reader is on the wrong screen, has already tried this twice, or is looking at an error it has never heard of.
The moment somebody has a question the tour did not anticipate, the tour is over and they are back where they started — except now they have also lost four minutes.
Day one, or on a flag your team sets. Both are guesses made before the thing that goes wrong has gone wrong.
Every tour is a set of steps pinned to a UI. Move the UI and the tour is either broken or, worse, quietly pointing at the wrong thing until somebody reports it.
Three properties that follow from not being a script, each of which is a page of its own.
Not because somebody scheduled it for day forty, but because it is not scheduled at all. It is available on the screen they are stuck on, at the hour they are stuck on it.
read the page →02Held against the account rather than the session, so the thing they never came back to in March is still known in May — including by the person who joined last week.
read the page →03The question a script could not anticipate is the entire category of question worth having a system for. Anything it cannot support, it hands to a person with the context attached.
read the page →A tour is the company talking on the company’s schedule. This is somebody available on theirs — which is the same difference as between a welcome email and a colleague at the next desk.
Announcing something to people who already use the product is a broadcast, and a broadcast is exactly what a tour is. Three cases where it stays the right tool.
You changed the navigation and every existing user needs to be told once. Nobody has a question yet, so there is nothing to answer — there is only news, and news is a broadcast.
A feature launch that existing users would not otherwise notice. The success measure really is how many people saw it, which is the one case where completion rate is the right number.
A genuine setup checklist with a defined end. Worth keeping — and worth accepting that the people who skip it are not lost, they are busy, and something has to be there for them in six weeks.
Amvio has no production numbers to show you and will not invent any. These are checks you can run this week on the tooling you already have.
Almost all of them will be in the first session. Then look at when your support volume happens. If those two distributions do not overlap, the tour is not covering the problem.
Then ask what in your product will ever offer them step three again. In most products the honest answer is nothing, and nobody has been able to tell.
Not to a cohort — to one account. Whatever answers that question is the thing this page is about, and a tour is not eligible.
Every onboarding flow has one. The interesting question is not how to make people finish it — it is what your product does for them six weeks later, when it finally matters.
You arrived with one frame. The person you have to convince next is holding a different one.
all four, in one page →Deflection counts the people who gave up. Every design decision in a bot follows from the one number it is paid on.
read the comparison →A document knows what it says. It does not know where anything is in this account, or that the step it describes was renamed.
read the comparison →The honest one, and the strongest. A person is better than this at four things. They are also asleep, ramping, and outnumbered.
read the comparison →Your users are already having the conversation. Right now it is with nobody.