For business owners and managers who have been offered a paid discovery phase before a digital or AI project and want to know whether it’s worth it. By the end you should know what happens in one, what you should get, and when you can skip it.
A discovery phase is a short, fixed piece of work before a build, in which a team investigates the problem, the people involved, the data and the constraints, and decides what is worth building. It ends with a report, a ranked list of opportunities and a build plan, so you can approve, change or stop the project with evidence.
In brief
- Discovery happens before anything is built. Its job is to make sure the build solves the right problem.
- A good discovery phase ends with documents you keep: a findings report, a ranked list of opportunities and a build plan, whoever builds it.
- UK government projects typically spend four to eight weeks in discovery. Commercial projects for small and mid-sized businesses can often be done in two to five days.
- Discovery is worth paying for when the project is large, the problem is unclear, or several people and systems are involved.
- A discovery that recommends not building something has done its job.
What happens in a discovery phase?
A discovery phase is mostly structured conversations, observation and analysis. The team talks to the people who do the work, looks at the systems and data involved, and turns what they find into a clear picture of where change would help most.
The idea is well established. The GOV.UK Service Manual describes discovery as the stage where you understand the problem that needs to be solved, and says you should not start building during it. The Design Council’s Double Diamond model, published in 2005, puts the same idea at the start of its design process: discover and define the problem before developing and delivering a solution.
Here is how a five-day discovery sprint typically runs for a business of 20 to 200 people:
| Day | What happens |
|---|---|
| Before day 1 | You send an organisation chart, a list of systems, documents for the three to six processes that matter most, and anonymised sample data |
| Day 1 | Kick-off with the owner or sponsor: goals, what success looks like, what must not change |
| Days 2 to 3 | Interviews with five to eight people in different roles, plus walk-throughs of the main processes and systems |
| Day 4 | Mapping processes, data and constraints; scoring each opportunity by the value it carries |
| Day 5 | Playback of findings, the ranked opportunities and a proposed build plan, with time for questions |
A shorter sprint of two or three days covers the same ground in less depth, usually for a narrower question.
What should you get at the end of a discovery phase?
You should get documents you can act on and keep, even if you choose someone else to build. A discovery that ends with only a slide deck and a proposal for the next phase hasn’t given you much to act on.
Our Discovery sprints produce:
- A findings report in plain English: what we heard, where time and money are lost, and what the constraints are
- Process maps for the main workflows, showing the steps, handoffs, volumes and failure points
- An inventory of systems and data: where information lives, its quality and who owns it
- A constraints log covering regulation, data protection, contracts and internal policies
- A ranked list of opportunities, scored by the value each carries and how hard it would be
- A build plan for the top one to three opportunities, with scope, approach, rough effort and the risks to manage
- A list of open questions, so nothing is assumed
A good test of the whole discovery is whether any competent team could start building from the plan.
How long should a discovery phase take?
As long as the question needs and no longer. The GOV.UK Service Manual says around 4 to 8 weeks is typical for government services, which often involve millions of users and complex policy.
Most small and mid-sized business projects don’t need that long. A focused question, such as “where should we use AI in our client onboarding?”, can usually be answered in two to five days if the right people are available and the documents arrive before day one.
A longer discovery makes sense when:
- Several departments or sites are involved, each with different processes
- The project depends on systems run by suppliers who need to be consulted
- You’re building a new product and need to talk to potential customers, not just staff
- Regulation is complex enough that compliance needs its own review
How much does a discovery phase cost?
Costs vary with the length and the number of people involved. Our Discovery sprints are fixed price, between £2,000 and £5,000 depending on whether they run for two, three or five days.
Fixed pricing matters here. A discovery billed by the day has no natural end, and the incentive is to keep discovering. A fixed price with a fixed set of outputs keeps both sides focused on answering the question.
Is a discovery phase worth paying for?
Usually, if the project that follows is significant. A discovery costing a few thousand pounds is a small share of a build costing tens of thousands, and the most expensive mistake in any project is building the wrong thing well.
It’s especially worth it when:
- You know something needs to change but not exactly what
- Different people in the business disagree about the priority
- The project involves personal data, regulated activity or several systems
- You want quotes from several suppliers and need a clear brief to compare them against
You can often skip it when the job is small and well defined, such as a new page on an existing website or a change to a single form. In that case, a short scoping conversation is enough.
How do you prepare for a discovery phase?
The best preparation is getting documents and diaries sorted before day one. Discovery time is short, so every hour spent chasing a login or rescheduling an interview comes out of the analysis.
Before a discovery starts, gather:
- An organisation chart and a list of who does what
- A list of the systems you use and who manages each one
- Any written procedures for the three to six processes that matter most
- Sample data, anonymised where it contains personal information
- Your data protection and information security policies
- A list of current suppliers and contracts that affect the project
- Recent management reports that show volumes and performance
Then book the interviews. Five to eight people across different roles usually give a full picture, and front-line staff often know more about where time is lost than their managers do.
Is a discovery phase the same as product discovery?
They’re related but different. Product discovery is an ongoing practice inside product teams: continually researching what users need and testing ideas before building them. A discovery phase is a one-off, time-boxed piece of work at the start of a project, often done by an outside team.
Most articles about product discovery are written for in-house product teams at software companies. If you’re a business deciding whether to commission a project, the discovery phase is the version that applies to you.
What’s different for regulated businesses in the Isle of Man and Channel Islands?
Regulated businesses need the constraints work done properly, because it decides what can be built. A discovery for a financial services, trust or legal business should cover the relevant regulator’s expectations, island data protection law and any client commitments about where data is stored.
That review often changes the answer. An AI tool that is ideal for a retailer may be unsuitable for a fund administrator because of where it processes data. Finding that out in discovery costs far less than finding it out after the build. Our guide to sovereign AI covers the data options in more detail.
Frequently asked questions
Do we own the discovery report?
You should. Ask before you start whether the report and build plan are yours to keep and share with other suppliers. Our Discovery outputs are written so the client keeps them whoever builds the project. A discovery whose findings can only be used by the team that ran it is closer to a paid sales process than an independent piece of work.
What happens if the discovery says not to build?
Then it has saved you the cost of a project that wouldn’t have paid back. A good discovery report explains why, and usually points to something smaller or cheaper that would help instead. It’s worth agreeing at the start that “don’t build” is an acceptable outcome, so nobody feels pressure to find a project that isn’t there.
Who from our business needs to be involved?
Plan for one sponsor who can make decisions, five to eight people across the roles that do the work, whoever manages your systems and data, and someone responsible for compliance if you’re regulated. Most people need an hour or two. The sponsor needs a little more time, mainly at the start and at the playback on the last day.
Is a discovery phase the same as a design sprint?
Not quite. A design sprint, the five-day method popularised by Google Ventures, focuses on designing and testing a prototype of one idea with users. A discovery phase steps back further: it works out which problems are worth solving in the first place, and produces a plan for the build. Some projects use both, with discovery first.
Can we use the discovery to get quotes from other suppliers?
Yes, and a good discovery should make that easier. A clear build plan with scope, constraints and priorities gives every supplier the same brief, so their quotes are comparable. Without one, each supplier fills the gaps with their own assumptions, and the quotes end up describing different projects.
How Hyperlabs can help
Hyperlabs Accelerate runs fixed or phased projects that change how a business works through digital, AI and go-to-market. Accelerate projects usually start with a Discovery sprint of two to five days, priced at £2,000 to £5,000, which ends with a findings report, ranked opportunities and a build plan you keep. For an overview of AI projects in particular, see our AI implementation guide.