What good service levels look like, and how to set them

The common failure is a service level report that is green every month while the business is visibly unhappy. That is not dishonesty. It usually means the measure describes something other than the experience.

Measure resolution, not just response

Response time — how long until someone acknowledges — is easy to measure and easy to meet, and nobody cares about it. What people experience is how long until they can work again.

Response targets are worth keeping as an internal signal. They should not be the headline, and they should never be the only thing reported to the business.

Set targets from the consequence of waiting

Work backwards from what the delay costs. A field crew that cannot submit tickets is losing a day of work. A finance team locked out during month-end has a deadline with consequences attached. A single person unable to access a reference document does not.

Targets built this way vary by service and by time, which is more complicated and considerably more useful than one number applied to everything.

Do not commit to what you have not resourced

A target set to look reassuring, with no change in staffing or process to support it, is a commitment to failing publicly every month. That erodes more trust than an honest target that is met.

If the business wants a faster target, that is a legitimate conversation — about what it costs to deliver it. Have that conversation rather than accepting the number.

Cover the whole path, including the part you do not own

Many service levels stop at the boundary of what IT controls, which means the reported figure excludes waiting on a vendor, a third-party approval, or a business decision. The user experiences the whole path.

Measure the whole path and report the breakdown. It shows where the time actually goes, and it is usually the fastest way to get a supplier contract renegotiated.

Define what is excluded, in advance

Planned maintenance, changes outside the agreed window, issues caused by systems you do not manage. Agree these before the reporting starts, not during the first month when a number is being challenged.

Publish, including the misses

Service levels build credibility only if the reporting is honest and visible. A monthly summary that shows the misses, with the reason and what is being done, does more for the relationship than a green dashboard that nobody believes.

The trap worth naming

Meeting every target while satisfaction falls means you are measuring the wrong thing. When that happens, resist the instinct to add more measures. Go back and ask a handful of users what “working properly” would mean for them, and rebuild from the answers.