Response times: what is realistic for a small business web partner
Not everything needs an answer in an hour. The four severity levels worth defining, the response each deserves, and why a written table beats a promise of fast.
The short answer
“How fast do you respond” has no single answer, because a site that is down and a request to change a paragraph are not the same event. A serious partner answers with a table: four severity levels, the response each deserves, and the hours it applies. That table is worth more than any promise of fast, because it says what happens at 2 a.m. on a Sunday and what happens on a Tuesday afternoon, and it can be checked against the monthly report.
The four levels
| Level | Examples | Response | Hours | Resolution target |
|---|---|---|---|---|
| Critical | Site down, security incident, checkout broken, data exposed | Alerted within minutes, usually from monitoring before anyone calls | Alert day and night; a person within the agreed window | Restore service first, fix after; typically within hours |
| Important | Form not delivering, error for some visitors, certificate warning, significant slowdown | Within a few business hours | Extended business hours | Same or next business day |
| Standard | Content changes, small features, new page, an integration tweak | Acknowledged within a business day, scheduled with a date | Business hours | Within the monthly change budget |
| Planned | Ideas, larger improvements, the next project | Discussed at the monthly call, placed in the quarter | Monthly rhythm | Per plan |
Why the table matters more than the number
A promise of “fast response” means whatever the person saying it needs it to mean. A table means: if the site is down at night, monitoring alerts a named person at once and the table says when work starts; if you want a paragraph changed, you hear back today and it is done this week. Expectations match reality on both sides, and the monthly report shows whether the table was honoured.
One thing worth saying plainly, because this article asks you to demand something: the four levels above are what we think a maintenance agreement should contain, not a description of what ours already says. Ask us for the table, the same way this article tells you to ask anyone else.
What monitoring changes
Most critical issues on a well-run site are found by monitoring, not by clients: the certificate that failed to renew, the form that stopped delivering, the site that went down. The response starts before the phone would have rung, and the client learns about it from the report. That is the standard to expect: not that the partner answers your call fast, but that you rarely need to make it.
- Ask for the severity table in writing, with hours and named on-call arrangements for critical.
- Ask what is monitored and who is alerted, so critical issues are found without you.
- Ask how times are reported. Detected, acknowledged, resolved, per incident, in the monthly report.
- Agree what counts as critical for your business. A shop and a brochure site differ; the table should reflect yours.
What this means for you
Do not ask for fast. Ask for the table, the monitoring behind it and the report that proves it. A partner with all three responds to a site down in minutes because a monitor woke them, acknowledges your change request today, and puts your idea in next month’s plan. That is realistic, sustainable and, over a year, far better than a promise nobody can keep.
Frequently asked questions
Should we expect our partner to be available at night and weekends?
Detection, yes: monitoring runs continuously and raises an alert at any hour. A person responding is a separate question. We do not staff a night and weekend rota; a person responds within the window in the agreement, which today states one response time per plan and the hours it applies, without severity levels. If an outage costs you money the moment it happens, a shop in peak or a transaction system, a named person outside business hours is a fair thing to require, and it belongs in the contract, with us or with someone else. Ask us for the severity table as you would ask anyone.
What is a fair response time for a normal change request?
Acknowledged within a business day, scheduled into the queue with a date, done within the monthly change budget. The acknowledgement is what matters: you know it was received and when to expect it. Silence for a week is the real problem, not a change that takes four days.
How do we know the promised times are met?
The monthly report lists incidents with times: detected, acknowledged, resolved. Over a few months the pattern is visible. A partner who cannot produce those numbers is not measuring them.