Experiment Not Running

You published an experiment but visitors are not being bucketed into variations: the results page shows zero visitors, or you cannot see the variation when you visit the page yourself. This guide walks through every reason this happens and how to fix each one.

Symptoms

• The results page shows 0 visitors after several hours.
• You visit the target page but do not see any variation applied.
• Calling avsb.getVariation(42) in the console (with your Experiment ID) returns null.
• The variation CSS/JS does not appear to be injecting into the page.

Checklist

1. Check this badge first: only Running experiments bucket visitors.
2. URL targeting and its match type live here, on the Targeting step.
3. Audience conditions are on the same step as URL targeting, not a separate one.

For step 1, if the status badge says:

• Draft: the experiment has not been published yet. Click Publish experiment on the Step 5 (Review & Publish) page.
• Scheduled: the experiment will start automatically at the date and time you set. Wait for that time, or open the Step 5 (Review & Publish) page and click Edit schedule to change or cancel it.
• Paused: the experiment was running but has been manually paused. Click Resume to start it again.
• Completed: the experiment has been stopped. You would need to create a new experiment or contact support to reopen it.

For step 3, when checking targeting rules:

• Is there at least one URL rule configured? An experiment with no URL rules will never match any page.
• Does the URL rule actually match the page you are testing on? Test your rule by comparing it to the browser's address bar. For example, a rule of /pricing with Exact Match will only trigger on exactly /pricing, not /pricing/ (with a trailing slash) or /pricing?plan=pro (with a query string).
• Want to ignore query strings and trailing slashes? Use Path Match. It compares only the path, so /pricing matches both /pricing/ and /pricing?plan=pro. Substring Match also works here, but it can over-match: /pricing would also match /pricing-old.

For step 4, common audience-condition mistakes:

• A segment condition is set (e.g. "plan = premium") but the visitor does not have that segment set.
• A device condition excludes the device you are testing on (e.g., "device = desktop" but you are on mobile).

For step 5, checking traffic allocation:

1. Each variation needs traffic above 0% to receive any visitors.
2. This overall percentage is a second gate: a low value here excludes most visitors from the experiment entirely, even if the variations above are split normally.

For step 7, run this in the console to force a variation:

// Replace with your actual experiment ID and a variation IDconst success = avsb.forceVariation(42, 1487); // Experiment ID, then Variation IDconsole.log('Force result:', success);

Inspecting the datafile

The snippet downloads a JSON datafile containing all running experiment configurations for your project. You can inspect it directly:

1. Open DevTools → Network tab.
2. Reload the page.
3. Filter by "datafile" or look for a request ending in datafile.json, for example cdn.avsb.cloud/<your snippet key>/datafile.json.
4. Click the request and look at the Response tab. Search for your experiment ID. If it is not in the JSON, the experiment may not have been published yet, or the datafile is cached.