Discovery is a product, and you are buying it
A paid discovery phase is a legitimate and often excellent purchase. It converts a vague idea into something that can be priced accurately, and it surfaces the problems that would otherwise appear in month three.
The test of whether you got value is simple: could you take the output to a different supplier and get a comparable quote from it? If not, it was not discovery.
The five deliverables
- A written specification of the workflows, roles, rules and exceptions, in language you understand.
- An architecture outline — what gets built, what is bought, what it integrates with, and where the data lives.
- A risk register naming the things that could go wrong, with what would be done about each.
- A phased plan with what is in phase one and what deliberately waits.
- A fixed price for phase one, with the assumptions it rests on written down.
What good discovery does that you cannot
It talks to the people who actually do the work, not only the people who commissioned the project. That is where the exceptions live — the workaround nobody documented, the customer with special terms, the step that only happens at month end.
The most valuable hour of most discovery phases is spent watching someone do the job. Not asking them about it — watching. What people describe and what they do are reliably different, and the gap is where projects fail.
Cost and duration
| Project size | Discovery cost | Duration |
|---|---|---|
| £15k–£40k build | £1,500–£4,000 | 3–7 days |
| £40k–£100k build | £4,000–£10,000 | 1–3 weeks |
| £100k+ build | £10,000–£25,000 | 3–6 weeks |
Roughly 5–10% of the expected build cost is a reasonable benchmark. Much less and it is a sales meeting; much more and it is starting to be a project of its own.
Ask for the credit arrangement
Many suppliers credit some or all of the discovery fee against the build if you proceed. That is a fair arrangement and worth asking about.
What matters more is that the deliverables are yours to keep and use regardless of whether you proceed. Get that in the engagement letter, because it is the difference between buying an asset and buying a pitch.
Warning signs
- No time spent with the people who will use the system
- Output that is mostly a proposal for their services
- A price for the build that arrives without written assumptions
- No risk list, or a risk list with nothing uncomfortable on it
- Reluctance to let you take the specification elsewhere