Experiment does not appear
Find out why a visitor does not see a variation, and fix it step by step.
You started an experiment, but you (or your visitors) do not see a variation. This page walks through the checks in the order in which ABlyft itself makes its decisions. Work from top to bottom and stop at the first check that fails.
How ABlyft decides
For every visitor and every page view, the snippet runs through the checks in the order described in Audiences & delivery. Developers find the exact sequence in the snippet sequence. If one check fails, the visitor sees the original page. Visitors are only assigned once, see The visitor was already assigned.
1. The snippet does not run
Open your page with ?ablyft_check=true at the end of the URL. If you do not see a pop-up with "ABlyft is
implemented", the snippet is missing on this page. Follow Embed the snippet.
If the pop-up appears but nothing else happens, the snippet may have stopped on purpose. The usual reasons:
- the page is in an iFrame or the visitor is a bot,
- you opted out with
?ablyft_opt_out=true(opt back in with?ablyft_opt_out=false), see Opt out, - Do Not Track or a prerequisite rule, for example a consent rule, applies,
- the Visual Editor is open, so the page is in editing mode.
The full list with console messages and fixes is in Common issues.
Testers often opt out by accident
If you tested the opt-out once, you stay excluded in this browser until you opt in again. Try a private window or another browser to compare.
2. The status is not Running
Only Running experiments are delivered to every visitor. Draft, Paused and finished experiments are not delivered at all. An experiment in Preview is only visible with a preview link. See the statuses and Preview & QA.
If you used a schedule, check that the start time has passed, see Scheduling.
3. Your changes are not published yet
ABlyft publishes the snippet automatically after you change something. Wait until the Publishing … indicator disappears. See Publishing delay.
4. No page matches
An experiment only runs on pages that are assigned to it. Check the URL targeting (including www, trailing slashes and
the protocol), the trigger and the JavaScript rule of the page. On single-page applications, use a URL Change
trigger. See Pages and
SPA navigation.
The example URL does not limit the experiment
The Example URL of an experiment is only used for the preview and the Visual Editor. Where the experiment runs is defined only by its pages.
5. The environment does not match
If the experiment is assigned to an environment, it only runs where the environment's rule applies, for example on your live domain but not on staging. See Environments.
6. The audience does not match
Check each audience rule against what you actually have in your browser, for example a returning instead of a new visitor. See Audiences.
7. An exclusion group removed the visitor
A visitor can only take part in one experiment of an exclusion group. See Exclusion groups.
8. Traffic allocation
With an overall traffic allocation of 50 %, half of your visitors never see the experiment, and that is normal. See Traffic allocation.
The visitor was already assigned
ABlyft stores a visitor's assignment in the browser (see Storage & privacy). If you were excluded earlier (by traffic allocation or an exclusion group), you stay excluded for this experiment, even after you fix the cause. If you saw variation A, you keep seeing variation A.
To test again:
- Open the page in a private window, or delete the website data of the browser.
- Or use the preview link of the experiment, which lets you pick a variation directly. See Preview & QA.
Consent can prevent saving
If your project only allows tracking after consent, a visitor may get a new decision on every page until consent is given. See Consent blocks tracking.
Still stuck?
Turn on the debug mode of your project. The browser console then explains for each experiment which gate failed. Your developer finds the steps in Verify the integration. If you still cannot find the cause, contact ABlyft support and send the URL, the experiment and what you see in the console.
Next steps
- Changes are missing: the experiment runs, but the page looks unchanged
- Debugging: console output and URL parameters