Define what the routine work covers
Agree on updates, backups, monitoring and routine fixes. List development separately: new pages, features and integrations. Regular maintenance does not mean unlimited development.
For external services (including hosting), define where responsibility ends. Your contractor can diagnose why the CRM is unavailable, but can't manage the CRM vendor's infrastructure. The plan should say how an incident like that gets handed over.
Response time is not resolution time
Response means the issue has been picked up and investigation has started. Resolution depends on the cause, the access available and any external systems. Don't merge the two into a promise of one deadline for every case.
Separate a full outage, a broken key feature and a cosmetic fix. Add working hours and a channel for urgent messages. Priority should be set by the impact on the business.
Keep a history and verify the result
Every task needs a URL, the symptom, steps to reproduce and the expected behavior. After the fix, test the affected user path and log the change. That helps when the problem comes back and when the work changes hands.
Agree on the scope before the first outage, not in the middle of it. What the routine work covers and how fast we respond is described on the technical support page. If your site keeps growing, website maintenance combines upkeep with a plan for new tasks.
