How to set a service level your team can actually meet
Set a response commitment from your own measured performance rather than your best week, and define the exclusions before a client finds them.
The promise you made before you measured anything
Somewhere in a proposal or on a website sits a sentence promising a response within twenty four hours. Nobody checked whether the team could do that. It was written because a competitor said it. Six months later the commitment is missed often enough that clients have stopped believing it and the team has stopped trying, which is the worse of the two outcomes.
A service level is not a marketing line. It is an operational commitment with staffing consequences. Setting one properly means finding out what you already do then promising slightly less than that.
What a service level excludes and what it costs
A service level cannot cover hours you do not staff. If nobody works Sunday then a Sunday clock is a promise to fail. It also cannot cover anything waiting on a third party. Payment gateway outages, a client's own approval, a supplier's dispatch and a hosting provider's downtime all sit outside your control and must be excluded in writing at the start rather than explained afterwards. Delays caused by missing information from the client belong in the same list.
The cost of a tighter commitment is cover. A promise that holds through leave and illness needs a second trained person on the queue, and anything extending past working hours means paid cover or a rota. Monitoring costs a little too. Somebody must track the clock and service fees in India carry GST at 18%.
Who should publish one and who should not
A two or three person business should not publish a service level. One person falling ill breaks it and the published promise then does more damage than silence would. Say instead that you usually reply the same working day and keep doing that. Reliability without a commitment beats a commitment without cover.
A business doing bespoke project work should not promise resolution times either. When each job is genuinely different the time to fix is unknowable at the point of promising, and a resolution commitment on unscoped work is a liability you have handed to your client. Commit to response and to a diagnosis update instead.
Publish one when volume is steady, when the work is repeatable, and when procurement at your clients asks for it. Support desks, hosting, maintenance retainers, logistics and payroll processing all fit. Publishing also helps when you are the more reliable supplier in a slow market and want that to be visible.
How to set the number
Measure before you promise. Pull the last three months of requests and record how long each took. If that data does not exist then start recording and wait a month. Guessing at this stage is how the original twenty four hour promise happened.
Separate response from resolution and commit to them differently. Response is acknowledging with a named owner and is almost entirely within your control. Resolution depends on the problem. Then define the clock precisely: the hours it runs, the days it does not, and what pauses it when you are waiting on the client.
Set the commitment at a level you already meet comfortably rather than at your best week. The temptation is to promise your average, which guarantees you miss roughly half the time. Add a named escalation path for the cases that will breach anyway. A client who is told early that something will be late stays calmer than one who discovers it.
How to judge whether you are meeting it
The average flatters badly. An average response time of four hours is entirely compatible with three clients waiting two days each, and those three are the ones who leave. Total tickets closed flatters for the same reason.
What matters is the breach count and the worst case. Report how many requests missed the commitment and how late the latest one was. Then look at the pattern. Breaches clustered on particular days or around one person's leave point at a staffing gap. Breaches spread evenly mean the commitment itself is set too tight for the team you have.
The common mistake
Promising resolution when you only control response. Sales agrees a four hour fix because the client asked for one. Delivery then owns a commitment whose timing depends on a vendor, a client approval or a problem nobody has seen before. This is expensive beyond the service credits. Teams working under an impossible clock begin closing tickets to stop the clock rather than to solve the problem, and you lose the measurement along with the service.
The milder version is setting the level from the competitor's website. Their staffing is not yours and you cannot see which of their promises they keep.
A note on putting it in a contract
A service level written into a client agreement is a contractual obligation and may carry credits, penalties or termination rights depending on how it is drafted. The exclusions, the measurement method and the remedy all matter in the wording rather than in your internal understanding of them. A qualified professional should review the clause before you sign it, particularly where a penalty is tied to a commitment you have not yet measured.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.