How to Build Your First In-App Product Tour with Appcues
Appcues is a no-code platform for building in-app tours, tooltips and announcements on top of an existing web app. It is free up to 100 monthly active users, with paid plans starting around $299/month, and publishing a flow does not require an engineering deploy.
What does Appcues need from your app first?
One thing: a way to identify users and read their attributes. Appcues does not need your source code, but it does need user_id plus whatever property you will target on — plan tier, signup date, or a feature flag.
Without that, you get a tour shown to everyone, and a tour shown to everyone is a banner with extra steps. Appcues is explicitly built for logged-in products; on a marketing site with no account model there is nothing to target against.
How do you install Appcues without a deploy?
- In Appcues, go to Settings → Install and copy the snippet.
- Add it to your app shell — the layout that wraps every authenticated page, not the login screen.
- Call
Appcues.identify(userId, properties)on login and on user-property change. - Send a test event from the Appcues debugger and confirm it arrives.
- Only publish flows after the debugger shows a resolved user, not an anonymous session.
If you run a single-page app, identify on every auth state change. A stale user object is the most common reason a tour fires for the account that signed out ten minutes ago.
How do you structure a tour that people finish?
Three steps, one goal. That is the whole rule. Appcues makes it easy to add a fourth and fifth step; resist, because completion rates fall off a cliff after three.
A structure that works for most SaaS dashboards:
- Step 1 — anchor a tooltip to the primary action button, one sentence on what it does.
- Step 2 — anchor to the result of that action (the created item), showing what success looks like.
- Step 3 — anchor to the next step the user would naturally take, with a single link.
Skip modal welcome screens. They get dismissed before the first step loads.
Which audience rules actually matter?
Appcues lets you target on events, properties, page URL and user state. Three rules cover most onboarding needs:
| Rule | Why it matters |
|---|---|
| Page URL match | Prevents the tour firing on settings or billing pages |
User property (e.g. plan = trial) |
Separates trial from paid onboarding |
| Event-based trigger | Show the tour after the user completes a prerequisite action |
Also set frequency to once per user, ever. Appcues defaults to re-showing on later sessions, which trains dismissal.
How do you test two variants properly?
Appcues includes flow-level A/B testing, and it is one of the few things the tool does better than free alternatives. The setup that gives a usable answer: two modal copy variants, a single success metric (first key action within 7 days), and a sample of at least a few hundred users per arm.
Below that sample, you will read noise as signal. Most teams run the test for three days and call it — resist that too, or the result will flip on you next month.
If you need the analytics side, pair Appcues with PostHog funnels rather than buying a second engagement platform.
Where does Appcues stop being enough?
At analytics depth. Appcues shows you who saw a flow and whether they completed it; it does not explain retention, paths or feature adoption across cohorts. Teams buying it as an analytics product will be disappointed.
Sequencing across flows is also weaker than Userpilot for multi-step onboarding. If onboarding is a five-stage journey with branching, you will feel the ceiling quickly.
And $299/month at the entry tier is real money for a feature that a competent engineer could approximate in a week — the value is in not maintaining it yourself.
Key Takeaways
- Appcues is free to 100 MAU, paid from about $299/month, and publishes flows without a deploy.
- Identify users with
Appcues.identify()on every auth state change; anonymous sessions break targeting. - Keep tours to three steps — completion drops sharply beyond that.
- Target by page URL, plan property and prerequisite event; set frequency to once ever.
- Flow-level A/B testing is its strongest feature; analytics depth is its weakest.
