A new CIO asks for a view of "what IT actually supports". You have three weeks. What do you produce and what do you avoid?

Produce a capability map with one thing overlaid

Two levels of business capability — perhaps forty boxes — with applications mapped onto them and a single overlay chosen for the question being asked. For a new CIO the most useful first overlay is duplication: how many applications support each capability.

A capability map alone is a taxonomy. It becomes a decision tool the moment something is coloured onto it, and duplication is the overlay that produces the most immediate conversation.

Why capabilities rather than the org chart or a process model

Capabilities are stable: they do not change when departments are reorganised, when a system is replaced, or when a process is redesigned. That stability is what makes them a coordinate system you can hang cost, risk, maturity and system count on for years.

The two things to avoid, both of which are common: drawing the org chart in disguise — if the boxes match department names, it will be obsolete at the next reorganisation — and drawing the value stream, since sequence and flow belong to a different artifact. There are no arrows on a capability map, and their absence is the fastest way to tell it apart from a process model.

Three weeks is enough if you scope it correctly

Levels one and two only. Level three exists in an appendix if you have it, and going to level four takes months and nobody reads it.

Source the application mapping from whatever inventory exists, accept that it is imperfect, and mark confidence rather than delaying for accuracy. A map with stated uncertainty delivered in three weeks is worth more than a precise one in three months.

What the conversation will be

Duplication in one or two capabilities — usually customer data, reporting or document management — where four systems do the same job for different regions or business units, each with its own contract and its own team.

Have the second overlay ready for that conversation: annual run cost per application, as bubble size. That is what turns an observation into a decision, and it is when the map stops being an architecture artifact and becomes a management one.