Internal IT with no contractual SLAs: how to measure anything

Without penalty clauses there is no external pressure and usually no useful measurement either. Here is how to put a number on internal IT that somebody will act on.

An IT service desk serving an outside customer has penalties in a contract. Miss the response time and it costs money, so somebody watches it.

Internal IT usually has none of that. There is a target, everybody knows the target is aspirational, and nothing happens either way. So measurement becomes reporting: numbers produced monthly, read by nobody, acted on never.

The mistake is to conclude that internal IT cannot be measured usefully. It can. It just has to be measured in money rather than in compliance.

Why the usual measures do not work

Response and resolution times tell you whether a promise nobody enforces was kept. When it was not, the answer is a conversation about whether the target was realistic.

Ticket volume goes up when things break and also when people start trusting you enough to ask. It is two different stories in one number.

Satisfaction scores are answered by the small group of people who answer surveys, and mostly measure whether the last person was polite.

None of these are wrong. They are just not decisions.

Measure the cost of friction instead

Nothing has a penalty clause, but everything has a cost, because everything consumes somebody's time and time has a rate.

Set the rate first. Engineer salary plus employer costs, over working hours. One number, agreed once, written down. Everything else follows from it, and agreeing it is a five minute conversation that makes every later conversation shorter.

Then price the friction you can see:

Each of those is countable from a ticket system, and each turns into pounds with one multiplication.

What changes when you do

A conversation about whether resolution times are acceptable has no natural end. A conversation about £96,000 a year in re-triage has one: either it is worth fixing or it is not, and the person who has to decide has what they need.

It also changes what you can ask for. "We need another engineer because we are busy" is weak. "These three categories cost £140,000 a year in avoidable handling, and here is the arithmetic" is an argument, and it might not end in another engineer, which is the point.

What not to do with it

Do not use it on people. The moment friction cost is read as individual performance, the numbers stop being true: tickets get closed early, reassignments get avoided by sitting on things. You will have destroyed the measurement to make a point.

It is a measure of the system, and almost every cause is a system cause: a form asking the wrong question, a boundary nobody settled, a routing rule written for a team that no longer exists.

Where to start

One category. The biggest one, for ninety days. Count it, price it, show the arithmetic, and take it to whoever can change something.

If that produces a decision, do the next one. If it does not, the problem is not measurement, and more numbers will not help.


This is what the [ITSM Financial Command Center](https://itsmdashboard.dev) was built to do: price a live queue rather than count it, with the arithmetic beside every figure.

This is part of what we do under Operation review. See the services, or tell us what is not working.

More like this