Discovery is the line on a quote that looks like consultants charging you to think. It is also, consistently, the cheapest insurance available on a software project.
It is a fair objection though, so here is what it should actually produce. If a firm cannot tell you this, the objection is warranted.
A decision log, not a document
The output that matters is not a deck. It is a written record of the decisions made and, crucially, the options rejected and why.
Six months in, someone will ask why the system works this way. Without that record the answer is a guess, and the team relitigates a decision that was already made properly once.
The data, actually looked at
Not "we will migrate your customer records" but: how many records, how many duplicates, which fields are empty, which contain something other than what the column name suggests.
This is the single most common source of mid-project surprises, and it is entirely knowable in advance by someone spending a day with an export.
Integration reality
For every external system: what it actually exposes, whether credentials exist, whether anyone still owns it.
- A documented API with a sandbox — days of work.
- A legacy SOAP service or nightly file drop — weeks, and permanently more fragile.
- A system whose vendor wants a five-figure fee to enable API access — a commercial problem, not a technical one, and better found now.
A scope with an explicit outside
The valuable half of a scope document is the part listing what is not included. "Reporting dashboard — version two" prevents an argument in month three that otherwise costs a fortnight and some goodwill.
A plan you can cut
Work ordered so that stopping early still leaves something useful. If the first thing built is the admin panel, cancelling at 60% leaves you with nothing. If it is the core journey, you have a product.
How long, and what it should cost
For a mid-sized project, one to three weeks and roughly a tenth of the build. If someone quotes you discovery at a third of the project, that is a different engagement wearing the word.
If you would rather do the thinking yourself first, our project brief builder walks the same questions and produces a document you can hand to anyone.
You can skip discovery. You will simply do it anyway, during the build, at the hourly rate, with code already written against the wrong assumptions.
What should a discovery phase deliver?
A written record of decisions made and options rejected, the data actually looked at rather than assumed, the reality of every integration, a scope with an explicit outside, and a plan ordered so that stopping early still leaves something useful. If a supplier cannot describe those outputs, the objection to paying for discovery is warranted.
How long should discovery take and what should it cost?
For a mid-sized project, one to three weeks and roughly a tenth of the build cost. If someone quotes discovery at a third of the project, that is a different engagement wearing the word. You can skip it, but you will do it anyway during the build, at the hourly rate, with code already written against the wrong assumptions.
Enjoyed this? Let's work together.
Start a project