Troubleshoot a failed or misleading test
This guide explains how to investigate common problems with domain verification, scripts, session execution, results and notifications.
Verification, import and validation
| Symptom | What to check | Possible resolution |
|---|---|---|
| Domain will not verify | Compare hostname, issued value, public DNS or served HTML, redirects and HTTPS. | Follow verification and retry the check after the public value is correct. |
| Import changes the scenario | Compare intended actions, source and saved steps, especially await, assertions and waits. | Use Playwright compatibility and conversion; import the corrected short path and inspect it again. |
| Could not find element on page | Inspect the current label, selector and prior page state. | Fix a specific selector/state mismatch and validate the scenario. |
| Element was on the page but hidden, disabled or covered | Check overlays, consent, readiness and the element actually selected. | A stable selector and predictable page state can resolve the failure. Validation can confirm whether the change worked. |
| Page did not load in time | Check reachability, redirects and target logs. | Fix access or the page response; repeat a single validation. |
Starting and completing a run
| Symptom | What to check | Possible resolution |
|---|---|---|
| Run will not start | Read verification, failed-scenario, available-hours and capacity messages. | Correct the reported prerequisite; revalidate and run a small baseline. |
| Fewer starts than requested | Compare Completed sessions, failed outcomes and the Concurrent users chart with target access logs. Ask support to inspect exact starts and lateness; those diagnostics are not shown in the report. | Resolve access or problems starting sessions; rerun a small representative test before increasing traffic. |
| Run stops early | Read the exact stop reason, failed-session rate and hours left in Plans & billing. | Resolve the reported failure, or buy hours or turn on auto top-up, then try a small run. |
| Scenario ran out of time | Compare measured duration, session timeout and first incomplete action. | Investigate slow or stuck work; change timeout only with a reason and recheck capacity. |
| Sessions stopped reporting | Check the report’s failure details and any message about sessions that did not finish. | Support can investigate problems starting sessions or receiving their results. These failures alone do not establish the website’s capacity. |
Example: blocked traffic is not a capacity result
A test requests 600 sessions but the target logs show access challenges and the report has failed or missing work. Low achieved traffic does not prove that the website can handle 600 successful visits. Access logs and browser errors can help distinguish blocked visits from failed actions. Support can investigate sessions that did not start.
When bot protection blocks the scenario, access can be arranged with the site owner through a rule limited to the authorised test. For Cloudflare, the verified domain’s setup instructions describe the supported configuration, whose access token must remain private. Global firewall changes or CAPTCHA bypasses are not appropriate substitutes. Validation and a small run can then confirm that the approved access works.
Reports, usage and notifications
| Symptom | What to check | Next action |
|---|---|---|
| Missing or unexpectedly fast timings | Inspect failures, the page-load counts, client-side navigation and collection rules. Page-load counts are not per-metric sample counts; ask support if exact TTFB/LCP coverage is needed. | Use metrics and measurement; missing observations are not zeroes. Repeat only after the coverage issue is understood. |
| Slow timings | Check geography, release, cache and achieved workload against baseline. | Application logs from the same period can help identify the cause. A repeat run can confirm whether a fix helped. |
| Higher usage than estimated | Compare actual recorded session duration, pauses, failures and the estimate’s source. | Reconcile through usage and billing and reduce or revise the next workload. |
| No notification | Check Test run history first, then Notifications settings and Slack connection or webhook receiver. | A completed run may have a notification delivery problem. Checking the connection or receiver can avoid repeating the test unnecessarily. |
Getting help
Support can help investigate an unresolved issue. Useful details include the run and scenario identifiers, timestamp, relevant settings, exact error and what you expected to happen. Any screenshots or excerpts should have passwords, tokens, payment details and personal data removed.