Your experiment is in the Running state but real visitors are still seeing the control — or worse, you can't tell whether the variant is being applied at all. Here's how to diagnose.
1. Check your targeting
The most common cause is overly tight targeting. If your URL rule is exact = https://store.example.com/checkout but the actual checkout URL is https://store.example.com/checkout/ (trailing slash) or https://store.example.com/checkout?step=1, no visitor matches.
Fix: switch to starts_with or contains. Targeting patterns guide.
2. Preview the variant manually
From the experiment editor, click Preview. This opens a temporary URL that forces the variant for your session. If the variant looks broken or unchanged in preview, the issue is the variant itself, not targeting.
3. Check that you're not the exception
Each visitor gets a stable hash that determines their bucket. If you're the one visitor checking, you might be in the control group consistently. Open the page in a different browser or incognito to test the other variant.
You can also use the preview link from the experiment editor to force a specific variant for testing.
4. Verify the goal is firing
If the experiment is running but conversion data is zero, the goal might not be firing. From the experiment results, check the conversion count. Zero conversions for 24+ hours usually means:
- The CSS selector for a click goal isn't matching anything (theme changed, classes renamed).
- A pageview goal is pointing at a URL that doesn't exist or isn't reachable from the variant flow.
- A custom event goal is firing in code that's broken.
5. Look for JS errors on the variant
If the variant has custom JS or CSS that throws an error, the visitor's tracker may stop firing — and you'll see no traffic on either control or variant. Open DevTools → Console while on the variant. Any red errors? Fix them, restart the experiment.
What it looks like when it's working
You should see traffic split roughly evenly between control and variants in the experiment dashboard within an hour of starting. Conversions take longer (depending on your funnel) but the visitor counts come fast. If the visitor count for any variant is zero after a few hours of decent traffic, something's wrong with targeting or assignment.