We were wrong about discovery
We used to sell a discovery phase that ended when the document was finished. It should end when a decision is made, and those are not the same day.
Discovery is worth paying for and most of it is sold at the wrong length. It should end when the next decision can be made confidently, which is usually a fortnight, not when a specification long enough to justify its fee has been produced.
What we used to do
Weeks of workshops, a long specification, a diagram of everything. The document was thorough and the client read the first section. By the time building started, parts of it were already being renegotiated, because decisions had been documented rather than made.
The error was not doing discovery. It was measuring it by the size of what it produced.
What changed
Discovery now ends at a decision: what is being built first, what is explicitly not in it, what it will cost, and what would make us stop. That can usually be reached in one to two weeks, and it is a shorter document than the one it replaced.
The rest of the detail is produced as the work reaches it, when it is informed by what the previous fortnight taught rather than by what everybody guessed at the start.
Why paying for it still matters
Free discovery is a sales activity. It is bounded by how much a supplier will invest before a contract, and it produces the answer that leads to a contract, which is not always the answer that a project should be scoped smaller or not started.
Paying for it buys a different thing: the right to be told the project should not happen. A discovery you paid for can conclude with a recommendation to buy a product instead, and that is a good outcome for the money.
- A decision on what is built first and what is excluded
- A cost range with the uncertain parts named
- The integration risks, checked rather than assumed
- A stop condition: what would make this the wrong project
Where we still get this wrong
Discovery still runs long when the client has not settled an internal disagreement. No amount of workshop time resolves a question about who owns a process, and we have historically kept going rather than saying that plainly.
The better response is to stop, name the decision the business has to take, and pause until it is taken. Continuing to bill through an unresolved internal question is how discovery turns into a document nobody acts on.
What to ask a supplier offering discovery
Ask what the deliverable is and what decision it enables. If the answer is a specification, ask which decisions it settles and who signs them.
Then ask whether the discovery can conclude that you should not build. A supplier who has never reached that conclusion is running a sales process with an invoice attached.
What to take from this
- 01Discovery should end at a decision, not at a finished document
- 02One to two weeks is enough to decide what is built first and what is excluded
- 03Free discovery is bounded by the sale it is trying to close
- 04A discovery that can recommend not building is the one worth paying for
Ask the question directly
- Does a short discovery mean more surprises later?
- It means the surprises arrive while the budget can still absorb them. A long specification does not remove uncertainty, it records assumptions in more detail and postpones finding out which were wrong.
- Should discovery fees be credited against the build?
- It is a common offer and it quietly reintroduces the bias, because the fee is only recovered if you proceed. Pay for discovery as its own piece of work and keep the recommendation clean.
- Who should be in discovery from our side?
- Whoever can decide, plus one person who does the work daily. Discovery with only managers produces a description of the process as it is believed to run.
Related answers
Web development
New builds, rescues and re-platforms, measured against what the site earns rather than against how it looks. The brand system that carries it is built in the same engagement.
AI transformation
Most AI projects fail because they add a chat box next to the problem instead of removing the step that costs the hours. We start from the workflow, and from the systems and the data underneath it.
Bring the decision you are stuck on
A call, forty five minutes, no deck. We will tell you if we are the wrong firm.