A CIO asks for a capability map of the enterprise. How do you scope it so it is actually used?

What the interviewer is testing

Whether you can keep an enterprise architecture artifact from becoming a year-long modelling exercise nobody consults.

The scoping question

What decision is this map going to support? Ask it first, because the answer determines the depth and the sequence. Common answers:

• Where should we invest next year?
• Where do we have duplicate systems?
• What does this acquisition overlap with?
• Which capabilities are at risk from a technology being decommissioned?

Each needs a different overlay and a different depth, and building only to the depth that decision requires is what prevents the exercise from consuming a year.

The scoping rules

Two or three levels, no more. Beyond that it becomes process modelling and stalls. Level one is roughly a dozen; level two perhaps sixty.

Capabilities describe what the business does, not how or who. Not the org chart with different labels, and not the system landscape.

Mutually exclusive — a given ability appears once, which is where most of the useful argument happens.

Start with the area under question, not with the whole enterprise. A complete map of a single business unit that gets used is worth more than a partial map of everything that does not.

What makes it valuable

The overlays, not the map. Cost, system count, maturity, strategic importance, risk. A heat map showing that six systems implement the same capability across three business units, or that the strategic priority has no funded work against it, is what changes decisions — and it is what people remember.

What a strong answer adds

Its durability as the reason it is worth building at all: processes change, org structures change every reorganisation, systems are replaced — but a bank still needs to originate loans, assess credit risk and handle collections. That stability is what makes multi-year comparison possible, and it is the argument for anchoring portfolio analysis to capabilities rather than to systems.

Common weak answers

Modelling the whole enterprise to level four. Producing the map without overlays, which is a taxonomy rather than an analysis.