After publishing a change, add one row with when it was released, what changed, which pages were affected, and the verification result. You can keep this log in a spreadsheet or task tracker.
Suppose a developer changed the menu on 4 October at 11:00 MSK, the site owner found an incorrect link at 11:20, and it was fixed at 12:00. These times will help you reconstruct what happened. Record when the error was detected separately from when it occurred, if the latter is unknown.
Fill in the entry right after publication
| Field | Example |
|---|---|
| Publication | 04.10, 11:00 MSK, primary example.ru |
| Change | New “Delivery” item in the main menu |
| Addresses to check | Homepage, /catalog, /product/1 |
| Person responsible | Sergey, task No. 17 |
| Result | At 11:20, the link points to /old-delivery; fix requested |
| Follow-up | At 12:10, /delivery opens on all three pages |
A date may be enough for ordinary text changes. For a server setting or a brief outage, include the time and time zone. Record the release on the live site, not when development was completed on the person’s computer.
Note changes that affect verification
The log should include a new version release, a change to the primary address, redirect changes, bulk editing, and cache settings. Separately note changes to the region, search query set, and AI questions. These affect which data you compare.
Don’t turn the log into a full retelling of the work chat. Design discussions that did not result in a release can stay in the task. For tracking purposes, what matters is when the changed state became available to visitors or the checking service.
For example, write “from 5 October, we’re monitoring Tula instead of Moscow.” Without this note, a change in rankings could be incorrectly attributed to a page edit.
Check the addresses in the entry
“SEO fixed” is not useful for acceptance. Record an observable result: the page title was changed, the address response was fixed, or an incorrect link was removed. For bulk work, note the rule and a few URLs to check.
After release, open these addresses manually. Then compare them with a suitable new audit. If a page was outside the audit’s scope, the absence of an error in the results does not confirm that the issue was fixed on that page. If needed, run a separate check with a suitable scope.
See the post-release instructions for more details. Save the date and address for each follow-up result. In the monitoring dashboard, check whether the report completed after publication and covered these pages.
Use the log to investigate a recurring error
Find the last release before the problem was detected and open the recorded URLs. Compare the conditions: the same address, authorization status, page variant, and date. Then ask the developer to check the relevant change and the necessary log interval.
A timing match does not establish the cause. For example, a new audit might have crawled an old page with an error for the first time. In the task, write “noticed after the menu update” until the person investigating confirms a connection. For more on checking a finding, see the guide to verifying an audit finding.
Once a week, look for entries with a recorded publication but no follow-up result. Assign a check or note what is preventing one. Keep completed rows, with links, in your document’s archive.
When handing the site over to a colleague, ask them to reconstruct one piece of work from the log: find the original problem, publication, and verification. If you have to look up the time or URL in chat, fill in that specific field. Don’t expand the spreadsheet with information nobody will enter.