For each error, agree who reads the message, who investigates the cause, who approves the change, and who checks the result after it goes live. In a small team, one person can handle several of these tasks. Record their name and contact details, along with a backup for when they are unavailable.
Suppose the owner runs a service business website, a freelancer changes the code, and the hosting provider maintains the server. The owner reads an outage notification first. They check the URL in a browser and pass the time and result to whoever can restore the site.
Fill in the table with names
| Action | Who is responsible in this example | What to pass on |
|---|---|---|
| Read the message and open the URL | Owner | Time, address, browser message |
| Find the cause of the page error | Developer | Explanation and possible fix |
| Investigate a server outage | Hosting provider | Log check results |
| Approve an important change | Owner | What can be changed and when |
| Check the fix after it goes live | Owner with developer | Addresses and repeated actions |
Replace the roles with names and contact details. If a row has no assigned person, decide who will handle it before the next outage. The person listed may bring in a specialist, but is responsible for getting an answer.
What the owner decides
The website owner defines important user actions and operational constraints. They can explain which pages generate inquiries, when an advertising campaign starts, and which changes must not go live without approval. This information cannot be derived automatically from an audit score.
Specify any constraints: for example, do not disable the form during an advertising campaign or change a service URL without approval. To verify that the form works, the owner needs to confirm with a manager that a test submission was received; they can assign a specialist to check the server configuration.
What to give the developer
The developer is responsible for changing the implementation within the scope of the approved task. The task should include the address, the observed fact, the expected result, and the acceptance criteria. A link to a long report without a specific example often leads to repeated clarification and wasted time.
Do not ask them to “fix everything the service flagged.” Some findings relate to product decisions, access restrictions, or content. The person doing the work needs to know which changes are approved and which need clarification. It is useful to agree on the acceptance criteria before release to avoid disagreements afterward.
You can use the structure in the developer task template. For example: “The service link in the main menu goes to the old address; it needs to point to /services/repair; after release, we’ll check the homepage and two service pages.”
When to contact the hosting provider
Problems with certificates, domain settings, network availability, and the server may sometimes require help from the hosting provider or administrator. The website developer may not manage these settings. Record in advance who has the hosting agreement and permission to contact support.
When reporting a problem, include the time and time zone, the specific address, and the observed result. “The site is sometimes slow” does not identify the relevant time interval in the logs. If the problem was confirmed with only one measurement, say so; whether it occurs repeatedly needs to be checked separately.
Keep the support contact and agreement number somewhere the backup can find them. Assign one person to approve changes while the issue is being investigated: multiple people restarting things at the same time will make it harder to determine what happened.
Test the agreement on one task
Choose a current finding in the monitoring dashboard. Work through it using the table: who has read it, who it was passed to, who promised to respond, and who will repeat the check. If no one is waiting for the result, the task can get lost among chats.
For a finding about search or meta tags, involve someone who can explain the expected state of the page. The guide to audit priorities can help you choose the first tasks.
Before taking leave, ask the backup to show you the log, hosting contact, and open task. Check the notification settings and agree to review the dashboard. External messages depend on the channel being connected and available. When the next real alert comes in, ask the recipient whether they saw it.