Discovery is not a formality

There is a version of discovery that most organisations have experienced. It runs for two weeks, produces a slide deck of personas and a journey map, and is followed by a delivery plan that was substantially written before discovery began. Everyone involved understands that it is a formality. It is often described, without irony, as de-risking.

The reason this version persists is that it is cheap and it produces artefacts. The reason it does not work is that it never generates the one thing discovery exists to generate: a finding uncomfortable enough to change the plan.

What discovery is actually for

Discovery is the only phase of a project where changing your mind is inexpensive. Every week that passes after it, the cost of a reversal rises — through commitments made, code written, suppliers contracted and expectations set. A discovery that does not change anything has not saved money by being efficient. It has deferred the change to the point where it is expensive.

So the useful measure of a discovery phase is not how much was learned. It is how much was learned that somebody did not want to hear. If a discovery concludes by confirming the brief, one of two things is true: the brief was unusually well founded, or nobody looked hard enough.

Three findings worth paying for

In our experience, the findings that justify the phase tend to fall into three categories.

• The problem is real but has been located in the wrong place. A client asks for a customer portal because the call centre is overwhelmed; the calls turn out to be about a billing document nobody can read.
• The organisation does not agree with itself. Four teams use the same word to mean four different things, and every previous attempt has failed on that unresolved ambiguity rather than on execution.
• The constraint that matters is not the one in the brief. The brief says the platform is slow. The constraint is that only two people are permitted to publish, and both of them are also doing another job.

None of these are discoverable from a workshop. They come from watching people work, reading the support queue, and asking the same question of enough people that the inconsistencies become visible.

How to tell a real discovery from a ritual

A few practical signals. A real discovery talks to people who are not stakeholders — the operator, the assessor, the person on the phone at seven in the morning. It examines the system as it is used rather than as it is documented. It produces a written position that somebody senior could disagree with. And it reserves the right to conclude that the project should not proceed in the shape it was proposed.

If your discovery cannot conclude ‘do not build this’, it is not a discovery. It is a mobilisation phase with research on the front.

That last point is the one clients find hardest, and it is the one that determines the value of everything else. A supplier who is paid to build is structurally unable to recommend not building. This is not a question of integrity; it is a question of incentive. It is worth being deliberate about who you ask, and what they will earn depending on the answer.

What it costs

A serious discovery on a substantial platform is six to ten weeks. On a programme with a budget in the millions, that is a rounding error, and it is routinely the first thing cut. The reasoning is usually that the organisation already knows its own business, which is true, and that the knowledge is already assembled in one place, which is almost never true.

The question worth asking before compressing it is simple: what would we have to learn in the next six weeks that would change this plan? If the honest answer is ‘nothing could change it’, the plan is not a plan. It is a commitment, and discovery is being purchased to decorate it.