New users don't fail because your product is hard. They fail because they don't know which three things to do first. A checklist answers that, and shows progress as they work through it.
Each step has a title, a description and an estimated time. The estimate does real work — “2 min” turns a vague obligation into something someone will do now rather than later.
A step also carries an action: a button that opens a URL, optionally in a new tab, or triggers something inside your product by name. The in-product action is the better one where it exists, since it drops people exactly where the step happens.
You can have the messenger close when the action runs, so people get on with the task instead of returning to a list.
Two options, and this is the decision that makes or breaks a checklist.
Automatically, by rules — the step completes when the thing is genuinely done, based on what you know about the user.
Manually — the user ticks it themselves.
Use automatic wherever you can detect completion. A checklist that stays unticked after someone has done the work is actively irritating — it tells them your product isn't paying attention.
Manual is the honest fallback for steps you can't observe, like “invite your team” when the invitation hasn't been accepted yet.
Finishing shows a success screen, with confetti if you want it. Worth having: completing something should feel like completing something, and a list that just goes quiet feels like it didn't matter.
Three to five, and only the ones that lead somewhere. The test isn't what your product can do — it's what someone must have done before your product is useful to them.
For most products that's a short list: connect the thing, invite the person, do the task once. “Read our documentation” and “Follow us” are steps that serve you rather than the user, and their presence makes the whole list feel like marketing.
