Fixed price has a bad reputation in software, and mostly it deserves it. It goes wrong when the price is fixed before anyone understands the problem. Then the studio pads the estimate, the client feels overcharged, and every change becomes a negotiation.
Discovery is the product
We do not quote a build until we have run discovery, and discovery is paid work with its own deliverable: a written scope, a set of flows, an architecture sketch, and a price that does not move.
A week of questions is cheap compared to three months of building the wrong thing.
What the scope contains
- The process being replaced, described as it works today
- A list of what is in, and — more usefully — what is explicitly out
- The integrations, named, with their known unknowns
- The data model at the level of entities and their relationships
- A sprint-by-sprint sequence with a demo date on each
When we say no
Occasionally discovery shows the project should not be built: the off-the-shelf tool is good enough, or the underlying process needs fixing before software is pointed at it. We say so, and the engagement ends there.
That has cost us projects. It has also brought back roughly half of those clients within eighteen months, with a better-shaped problem.