What belongs in a service catalogue, and what does not

Most service catalogues are a list of everything IT owns, which is an inventory rather than a catalogue. Here is the difference, and how to write one people use.

Most service catalogues are a list of everything the IT department owns. That is an inventory. A catalogue is a list of what somebody can ask for, written in their words, and the difference decides whether anybody ever opens it.

The test for every line

Would somebody outside IT ask for this by name?

"A laptop for a new starter" passes. "Active Directory" does not: nobody outside IT wants Active Directory, they want to be able to log in. "Access to the finance system" passes. "SQL Server" does not.

That single test removes about half of most catalogues, and everything it removes is still real. It just belongs somewhere else, which is the next section.

Services and the things underneath them

Two lists, not one.

The catalogue is what people ask for. A dozen to thirty lines for most organisations, written in the words the person asking would use.

The service model underneath it is what each of those depends on: the systems, the suppliers, the infrastructure. That is where Active Directory lives, and it matters enormously, because it is how you answer "what breaks if this goes down".

Keeping them in one list satisfies nobody. The person asking cannot find what they want, and the person answering an outage question has to read past things that are not systems.

What each line needs

If a line cannot be given a truthful answer for all six, that is worth knowing before it goes in.

The things that quietly ruin one

Writing it for an audit. A catalogue written to prove the discipline exists is complete and unusable. Write it for the person who has just joined and needs a laptop.

Too many lines. Past about forty, nobody browses, so everybody rings the person they rang last time and the catalogue is decoration.

Times nobody meets. One over-promised line teaches people the whole document is optimistic, and then they ring somebody for everything.

Never revisiting it. A catalogue is out of date within a year. Twice a year, ask which lines nobody has requested and which requests never matched a line. Both tell you something: the first is dead weight, the second is what people actually need.

Where to start if you have nothing

Do not start with a blank page. Take the last three months of requests, group them by what was actually being asked for, and sort by how often. The top fifteen groups are your catalogue, in the words the people asking already used.

That takes an afternoon and it is more accurate than a month of workshops, because it is built from what people did rather than from what anybody remembers them doing.

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

More like this