Troubleshooting
The five-minute checklist for a test that seems dead — nothing triggers, no visitors, no conversions — and how to verify variants without skewing results.
Browse docs
A test that seems dead — nothing triggers, no visitors, no conversions — almost always has one of five causes, and none of them is a malfunction. Work through this checklist before anything else.
Start on the test's own page: while a test is running, a Launch status panel sits above the results and answers the first questions for you — snippet detected, target page visited, visitors assigned, conversions recorded. Each check below maps to one of its rows.
1. You're not on the page the test targets
A test only triggers on pages its URL targeting matches. A test targeting /page-a as a Specific page does nothing on your homepage, your thank-you page, or anywhere else — visiting the site is not enough, you have to open the target page itself. The Launch status panel names the exact URL when nobody has visited it since launch.
Match-type gotchas: a * only acts as a wildcard under Advanced pattern matching — under Specific page or Group of pages it's a literal character, and the wizard now refuses to save that combination. Advanced pattern URLs must include at least one *; the wizard rejects a wildcard pattern without one.
2. Traffic allocation excluded you
Below 100%, the allocation gate is deterministic per visitor while the allocation percentage stays unchanged: returning or reloading does not re-roll eligibility. Raising allocation later can admit previously excluded visitors. At 50% your own browser may be excluded, so either set allocation to 100% or use Preview on site, which shows any variant regardless of the gate.
3. You're in the control group
Visitors assigned to Control see the original page — that's the experiment working, not failing. If you open the target page and nothing changes, check the visitor count on the test page before concluding anything: if it went up, you were assigned. To test as a different visitor, close every private window before starting a new private session, or use a separate browser or browser profile. Private windows open at the same time share the same Optimo visitor cookie and assignment.
4. A redirect variant on a broad URL match
A redirect variant can fire on every page the test's URL match covers. Combined with a site-wide wildcard like https://example.com/*, it can redirect variant visitors away from matching pages — including the thank-you page where your goal lives, which can zero variant conversions. For a page-vs-page test, target the test page with a Specific page match; the wizard warns about this combination.
5. Conversions aren't counting
Conversions only count for visitors who are in the test — zero visitors always means zero conversions, so fix assignment first. For pageview goals, the goal URL is matched on origin plus pathname: use the address the browser actually shows, watching trailing slashes and www. And after any change to a running setup, hard-reload before verifying — browsers cache the Otter script for up to an hour.
Best practices
Verify variants with Preview on site, not by reloading and hoping — previews never touch your results, and its Goals panel lets you fire each supported on-site goal before launch.
Keep traffic allocation at 100% unless you deliberately need a holdout.
Test the full funnel once per variant: target page, variant experience, goal page.
When in doubt, the Launch status panel on the test page tells you which of these five is in play.
Frequently asked questions
Quick answers to the questions teams ask most about this part of Otter.