What a diagnostic should produce

Most diagnostic work produces a document. The document is thorough, it is usually accurate, and it changes very little. Everyone recognises the pattern: a considerable amount of interviewing and analysis, a presentation that is well received, and a business that continues in exactly the direction it was already travelling.

We think the failure is one of definition. A diagnostic is not an act of description. Its purpose is to put a small number of decisions within reach of the person who has to take them, and everything else it produces is overhead.

What we look at first

A diagnostic begins with four questions, asked in this order because each one constrains the next.

Where does the money actually go

Not the budget, but the cash. Which activities consume it, which return it, and over what period. This is frequently the first surprise, because reporting is organised around departments while cash is consumed by projects that cut across them.

What decisions are open

The list of things that have not been decided, separated carefully from the list of things that have been discussed. In most organisations these two lists have been quietly merged, which is why so many projects are waiting on a decision everyone believes has already been taken.

What depends on what

The dependency map is where delay actually lives. A project rarely fails because one task is slow. It fails because three tasks were sequenced in an order that put the slowest one last.

Who can say no

Formal authority and real authority are different, and the gap between them explains most of what looks like inertia. Anyone who has watched an approved initiative stall for a quarter has watched informal veto power at work.

Three outputs, and no others

A diagnostic should end with three things, all of them short.

• A statement of position. One page describing where the business or project genuinely stands, written plainly enough that someone who was not involved can understand it and someone who was involved cannot dispute it.
• A decision list. Between three and six decisions, each with an owner, a date, and the information required to take it. If the list is longer than six, the work is not finished, because prioritisation has been handed back to the client.
• A sequence. The order in which those decisions should be taken, with the reasoning. This is usually the most valuable page and the one that takes longest to produce.

What it should not contain

Three things tend to fill diagnostic reports and should be resisted.

Benchmarks that flatter. Comparison to a peer group is useful when it is uncomfortable and misleading when it is chosen to reassure. If the peer set was selected after the data was gathered, it is decoration.

Recommendations that could apply to any company in the sector. If a line could be moved into a competitor's report without alteration, it is not a finding. It is a description of the industry.

Anything that cannot be acted on within the next fortnight. Long horizon observations are not wrong, but they crowd out the small number of items that determine whether the next quarter is productive.

Why two to three weeks

The duration is not arbitrary. Below roughly two weeks, the work relies on what people say rather than on what the numbers and the calendar show, and interviews alone reproduce the organisation's existing account of itself. Beyond three or four weeks, two things happen: the diagnostic starts to be treated as the project, and the people who were candid in week one begin managing their answers.

There is also a practical reason. A short, sharp exercise preserves the client's willingness to run another one later. A long one tends to be the last diagnostic anybody agrees to.

The test

We hold ourselves to a single measure, and we suggest clients apply it to anyone doing this kind of work. A month after the diagnostic ends, can you point to a decision that was taken because of it, and would you have taken that decision anyway?

If the honest answer to the first half is no, the work produced a document. If the honest answer to the second half is yes, the work produced confirmation, which is worth something but is not what was commissioned.