Onboarding Patterns: Tours, Tooltips, Checklists, and Empty States
Onboarding patterns help a new user start without a mandatory tutorial. Compare product tours, tooltips, checklists, progressive disclosure, and sample data.
Why is one opening tour not enough?
The most common mistake is treating onboarding as a one-day event: a sequence of five screens shown at login that nobody sees again. Laws of UX argues that traditional tours can get in the way of active users, who want to start immediately. Its proposal is to use tooltips, which offer lightweight, contextually relevant help, and to reveal features gradually (Laws of UX: Onboarding for active users).
That argument rests on the “paradox of the active user,” attributed to Carroll and Rosson in 1987. It is a reasoning framework, not quantitative data: the article itself offers no figures.
How do the patterns compare?
| Pattern | What it does | When it works | Risk |
|---|---|---|---|
| Guided tour | Walks through the interface step by step | One prior decision must be made before the rest makes sense | Becomes an obstacle if long or impossible to skip |
| Tooltip | Explains an element in its context | A feature the person has just found | Piles up and covers the interface |
| Checklist | A list of setup tasks | Goal-oriented products | If it grows without pruning, nobody finishes it |
| Progressive disclosure | Shows the essentials and hides advanced options | Products with many features | Hiding something the person needs |
| Empty state | Explains what will appear and how to create it | Screens that start blank | Generic messages with no action |
| Sample data | Fills the screen with example content | Dashboards and CRMs where blankness confuses | Forgotten sample data left in place |
Appcues classifies these patterns by who drives the action: the product (tooltips, modals), the person (checklists, empty states), or an agent. Its practical test is that a modal should appear only when one decision has to happen before anything else makes sense (Appcues: Product tours: when to use which UI pattern). It is written by a vendor that sells these tools, so treat it as a rule of thumb, not as evidence.
What does progressive disclosure say?
NN/g defines it as showing only a few of the most important options first and revealing the rest on request. The article dates from 2006 but remains a basic reference. It warns that designs with more than two disclosure levels typically have low usability, and recommends deciding what goes on the initial level with task analysis, field studies and frequency-of-use data, supplemented by observational usability testing (NN/g: Progressive Disclosure). For onboarding, that means deciding what is essential for the first result and leaving the rest for later.
How do you design a checklist people actually finish?
Appcues describes a finite list of setup steps with visible progress, and recommends that each item produce a tangible outcome the user can see. It warns that checklists fail when every new feature adds an item until the list is twenty points deep, and advises pruning them regularly. As a practical rule, keep the steps few and review them whenever the product changes.
The key is that the list is not a disguised tutorial. If every step is “learn this,” people close it. If every step is “connect your data source,” the list pushes toward a real action.
Illustrative example
Illustrative example: an invoicing tool for freelancers opens with a nine-screen tour. Almost nobody finishes it. The team replaces it with a three-item checklist: connect a bank account, create the first client, and send a test invoice. The home screen, with no invoices yet, shows an empty state with a “Create my first invoice” button and a link to a sample invoice. Tooltips appear only when the person opens the tax module for the first time. Each pattern responds to a different situation.
If the opening moment is a blank screen, review empty states. And if the first result depends on a long step, the end of that step carries more weight than it seems: the article on the peak-end rule explains why memories of an experience concentrate on its peak and its close.
Common mistakes
- Building a mandatory, long tour with no option to skip it.
- Creating a tutorial mode separate from the real product, which nobody returns to.
- Putting a tooltip on every element until the interface is covered.
- Writing checklists whose tasks are lessons instead of results.
- Leaving sample data in place without a clear path to replace it with the person’s own.
- Measuring onboarding by how many screens were viewed instead of how many people reached the first result.
What you can do tomorrow
- Measure how many new people reach your first useful result, and at which step they drop off.
- Replace one tour with a three-task checklist that has tangible outcomes, and compare completion rates.
- Check your start flow for dark patterns, such as an exit button that is hidden. An onboarding that pushes is not onboarding; see dark patterns.