After you’re told “it’s done,” open the address from the original task and repeat the action. Then check that the new audit was run after publication and included that page. A report entry disappearing does not confirm a fix if the URL wasn’t crawled.
For example, a menu link led to /old-delivery and an error page. Check the menu link on the same pages, rather than only opening the new /delivery URL directly.
Find the original task
Find the task and report that the fix was based on. Note the URL, its previous behavior, and the agreed expected result. If the original task only named a type of error, first reconstruct a specific example.
For a group issue, select the same test pages. Don’t replace a complex URL with a simpler one just because it looks correct. Exceptions in a template often lead to false acceptance: the main page works, but a variant with an additional block still produces an error.
Write down the steps and expected result. If the task was a general request to “fix SEO,” identify the specific page and assignee. For help understanding the original issue, see the guide to audit findings.
Compare the publication and check times
Ask the person who made the change when it became available on the live site. A “task completed” timestamp may refer to a staging environment, a local version, or a change awaiting publication. For acceptance, what matters is the live version that the service checked.
Check the date of the repeat audit. If it started before publication, the old result cannot be used to disprove the fix. If publication took place during the crawl, some pages may have been checked in different states. In that case, repeat the crawl after publication is complete.
If the results differ, check the cache: a saved copy of the page may still show the previous version. Don’t change the acceptance criteria to get a green status. Record the time, URL, and response your browser received.
Make sure the right URL was checked
The absence of a message in the new report confirms a fix only if the audit covered the relevant pages and completed the check. If the URL wasn’t checked, its status remains unknown. This is especially noticeable on large sites with multiple templates.
The daily automated audit of a connected site crawls up to 100 pages. A separate full run is still subject to the limit of the selected plan. With an active paid subscription, included manual checks can be run without a monthly quota. The scope of the selected run is still limited by the plan.
Find the test URLs in the new technical report. If there isn’t enough information, repeat the appropriate check or action in a browser. For a template, use different examples: a standard card and a variant with additional fields.
Repeat the result of the fix directly
| What was fixed | What to repeat | What to record |
|---|---|---|
| A menu link | Click it from the original page | The destination URL and the page that opens |
| A card title | Check the page title | The new value and URL |
| A shared block | Open the agreed different pages | The list of examples checked |
| A form | Submit a clearly marked test | The time and confirmation of the final recipient |
A title is the page name set by a meta tag and usually visible in the browser tab. If the change concerned a link, an improvement in search ranking is a different result. Acceptance is based on what was agreed to change. See the guide to working with a contractor for more details.
Record the result and any remaining questions
For example: “The fix was published on 4 October at 12:00. At 12:15, the menu link opens /delivery from the homepage, category, and product page. The new audit at 12:30 checked these URLs, and the previous issue is no longer reported.” This conclusion does not automatically apply to other pages.
If the browser shows a problem but the report doesn’t, send the developer the original steps. If everything looks good when checked manually but the issue remains in the report, compare the dates, URL, authentication status, and any saved copies. Record both results until the discrepancy is resolved.
Later, check the monitoring dashboard to see whether the error has returned. Keep the record of the confirmed fix in the task: if the issue appears again, it will need a new timestamp and investigation, not the deletion of the old check.