Start with issues that prevent visitors from opening an important page or completing an intended action. Then address recurring problems and findings that still need verification. The number of tasks depends on the team: it is better to finish one important task than to start ten without completing any.
Use the latest report, the list of work already assigned, and an estimate of the time available from the person doing the work. For each new task, define what will count as completion.
Start with obstacles for visitors
First, check availability and the main user journey. If a key page does not open or a form does not accept an inquiry, there is little point in spending the whole week improving secondary metadata. However, a successful HTTP status from the website does not confirm that the form works: check it separately.
Describe the problem in terms of what the visitor is trying to do. Instead of “bad page,” write “the visitor cannot open the service page using the link in the menu.” This fact helps the developer identify what needs fixing and helps the owner understand the priority.
If the form does not deliver inquiries, confirm this with a test alongside the manager. If a service page does not open from the menu, save the link and the browser message. These tasks have a clear impact and do not require inventing a figure for lost sales.
Group similar symptoms
Take up to ten significant findings from current reports. For each one, add one or more URLs to check, the date, and the current result. Set aside errors that have not yet been reproduced: investigating them can also be a task for the week.
Group similar symptoms carefully. Several identical messages may have a common cause, but this needs to be checked. First, note the suspected scope—a template, a group of pages, or a single URL. The person doing the work can clarify the actual scope later.
For help choosing the first tasks, see the guide to prioritization. If there is a common issue with the menu or a template, save several URLs as one task. Ask the person doing the work whether they really share the same cause.
Compare candidates using one example
Suppose an online store has three issues:
| Candidate | What is known | Decision for the week |
|---|---|---|
| The delivery link leads to an error page | It recurs in the menu on the homepage and a product page | Fix it first and repeat the navigation check |
| Products have identical Titles | Confirmed on three product pages | Check the shared template and assess the change |
| There was a brief outage | The site responds now; the cause is unknown | Get the logs for the time of the event |
Title is the page name shown in a browser tab, set by the meta tag. For the second task, agree on a naming rule; for the third, the result for the week could be an explanation of the cause or a list of missing data. Do not set a deadline for fixing an unknown issue before it has been assessed.
Define how to verify that the work is complete
For each task, specify the initial state, the expected behavior, and how it will be checked. The URL must be exact, and the expected behavior must be agreed on. “It got better” cannot be assessed objectively. “The specified URL opens the correct page and does not lead to the wrong section” can be checked.
Decide separately who publishes the change and who verifies the result. If it is the same person, they still need to check at two points: after publishing and on the published version. A developer’s local result does not yet show the state of the live website.
For a link: “After publication, the menu link opens /delivery on three pages used for verification.” For an outage investigation: “The logs for 04.10 11:00–11:15 Moscow time were checked; the cause was described or the missing data source was identified.” A separate instruction can help verify a finding.
Check the results at the end of the week
Compare the new results in the monitoring dashboard with the tasks. The daily audit covers up to 100 pages: for a URL that has not been checked, open it manually or get a suitable separate report.
Instead of “worked on SEO,” write: the link was checked after publication; the naming rule was agreed on and is awaiting release; the hosting provider’s response is still pending. For each pending item, name a person and the date of the next follow-up.
If a new outage occurs, revise the plan and say which task is being postponed. Keep the postponed task with the reason rather than returning it to the general list without its history. At the next review, start with unfinished commitments and new facts that have changed the urgency.