What a discovery phase should actually produce
Discovery is often sold as a workshop and a slide deck. That is a sales artefact. A real discovery leaves you with documents you can build from, or decide not to build from.
The four artefacts
A scope with numbered flows and explicit exclusions. A data model sketch naming the entities, their relationships and the rules that govern them. A risk register covering integration, data quality and adoption. An estimate with assumptions written next to each number.
If any of these is missing, the phase produced enthusiasm rather than information.
What discovery should cost
Typically two to five percent of the build budget, or a fixed fee for a defined number of days. It should be credited against the build if you proceed, and it should be short enough that you feel the urgency.
The value of deciding not to build
A discovery that concludes the process should be simplified rather than automated has saved more money than most builds create. That outcome is a success, and it should be an explicitly allowed one.
Common questions
- Can we skip discovery to save time?
- You can, but the work does not disappear. It moves into the build, where changes are more expensive and estimates are less reliable.
- Who should attend discovery workshops?
- The people who do the work daily, a decision maker who can settle scope, and whoever owns the data. Absent decision makers are the most common cause of a second workshop.
Next step
If this question is live in your business right now, describe the situation and you will get a written view on the options, including whether you should build at all.
Discuss your projectRelated notes
All insightsCustom software or off-the-shelf: how to decide
A practical framework for choosing between configuring an existing product and commissioning custom software, based on where your process actually differs.
ReadStrategyHow to write a technical brief that gets accurate quotes
The brief structure that separates a project that can be estimated from one that cannot, with the eight sections worth writing before you talk to any developer.
ReadStrategyRebuild or repair: an honest test for an ageing website
How to tell whether your existing website needs a migration or a targeted fix, using five checks that take an afternoon rather than a month.
Read