What is a discovery sprint, and why is it needed?
A discovery sprint is a short, bounded period before design and development devoted to understanding the problem, users, and existing systems. English-language sources call it a discovery sprint or discovery phase. Its duration—usually two weeks—and deliverables are defined upfront.
Why is it needed? Because the details determining project duration and cost are found in daily work, not just the first meeting. “Let our dealers place orders” could mean a one-screen form or a platform with credit limits, promotion rules, and integrations across several systems. A timeline and budget offered before making that distinction is a guess, however well-intentioned. Discovery turns that guess into an informed range. That is also the short answer to “how does a software project start?”: by validating the problem rather than writing code.
Discovery is not performed to create an analysis report that sits on a shelf. Its purpose is decision-making: what the first release will and will not include, the sequence of work, and how success will be measured.
What happens in two weeks?
The two-week rhythm is simple: listen and observe in the first week, organize findings and decide in the second.
Week 1: stakeholders, fieldwork, and systems
- Stakeholder interviews. Decision-makers, process owners, and daily users are interviewed separately. This often reveals where management's objectives and field needs diverge.
- Field observation. Time is spent where work happens, alongside the person doing it. People omit exceptions when describing a process; observation reveals copied spreadsheets, orders confirmed by phone, and steps described as “we always fix that manually.”
- System and data inventory. Which systems are used, where does data originate, who updates it, which systems have APIs, and which can be reached only by file transfer? Integrations are among the least certain parts of a project; the inventory makes that uncertainty visible early.
Week 2: map, metrics, and scope
- Workflow map. The current process—steps, roles, tools, and bottlenecks—is collected into one map. The target workflow is drawn over that map.
- Success metrics. Document in measurable terms what should change when the product goes live, such as hours to complete an approval or steps to place an order. If the current value is unknown, how to measure it is also part of the output.
- First-release scope proposal. Select the workflows with the greatest impact on the metrics; list what is excluded from the first release and why.
- Roadmap and estimate range. Sequence the first release and subsequent stages, giving a recommended team and timeline range for each. The range makes remaining uncertainty visible rather than hiding it.
By the end of the second week, a clickable prototype of the main flow is also ready. Discovery defines what will be tested in it and with whom; testing with real users takes place in week 3.
What do you have at the end of discovery?
After two weeks, you should have more than a presentation: working documents that directly inform the next steps.
- Workflow map showing current and target processes
- User roles and the core needs of each role
- System and data inventory; integration methods and known risks
- Measurable success criteria and measurement methods
- First-release scope proposal and list of exclusions
- Clickable prototype of the main flow and plan for user testing in week 3
- Phased roadmap, recommended team, and timeline range
- Open questions, with an owner and deadline for each answer
These documents should be usable whichever team continues the project. Discovery's value is independent of who you work with afterwards.
A hypothetical example: cash management
Suppose a mid-sized company's finance team tracks cash across five separate banking portals and manually updated spreadsheets; payment approvals wait in email chains. The initial request is likely: “Let's see all accounts on one screen and forecast cash flow for 90 days.”
In the first week of such discovery, we observe the finance manager's morning in context: which portals are opened in what order, where balances are copied, and whose approval each payment needs. The inventory clarifies which banks offer APIs and whether due-date information for collections is held in ERP or spreadsheets. In week two, amount- and role-based approval rules are mapped; payment approval time might be selected as the success metric.
These findings can change scope. If due-date data is fragmented, a 90-day forecast will not produce reliable results in the first release. The initial scope then becomes account visibility and approvals; forecasting waits until data has been consolidated. The Corporate Cash Management product concept, using the fictitious X Ödeme, shows what such a product could look like.
When is discovery needed, and when is it not?
Not every job requires two weeks of discovery. The greater the uncertainty, the greater discovery's value.
| Situation | Is discovery needed? |
|---|---|
| A new platform for multiple teams | Yes. Each team experiences the process differently; scope cannot come from a single meeting. |
| A product replacing a legacy system | Yes. Undocumented rules emerge only when usage is observed. |
| Integration work connecting many systems | Yes. Data quality and access methods directly determine duration. |
| A small, well-defined addition to a known workflow | No. Users, data, and rules are already known. |
| An internal tool for one team with a clear flow | Usually not. After a short definition exercise, work can start directly with a prototype. |
For cases in between, discovery may take one week. The test is simple: if you can write the first-release scope on one page today in a form everyone involved will approve, lengthy discovery is unnecessary.
How do you prepare for discovery?
Discovery's productivity depends on who and what are available on day one. Preparation takes a few days and is largely scheduling work. Use the project brief tool to collect the problem, roles, systems, and success criteria on one page and share them before discovery.
Preparation checklist
- A decision-making sponsor and a product owner responsible for day-to-day work are identified.
- Interview time is reserved with at least one person from each daily user role.
- Place and time are arranged for observation: warehouse, call center, finance team, or field team.
- Current spreadsheets, forms, and screenshots are collected; personal data is masked.
- The system list, any API documentation, and technical owners' contact details are ready.
- Previously attempted solutions and reasons for abandoning them are noted.
- Known constraints are documented: legal requirements, security policies, budget cycles, and immovable dates.
Limiting interviews to managers is discovery's biggest risk. The people who know the process best are often those doing it every day; the schedule must include them.
What happens after discovery?
At the end of discovery, inputs for the next stage are ready. Custom software development then continues as follows in our process:
- Definition and prototype (2–3 weeks). The discovery prototype is tested with real users in week 3; feedback clarifies screens for critical flows. The first-release scope is finalized together after these tests.
- Build (2-week sprints). Working software is demonstrated every week. The jointly scoped first release targets go-live in weeks 6–12.
- Grow and operate. Success metrics are tracked in production and roadmap modules are added in sequence. Large programs progress in stages rather than all at once: the first is planned to go live in 6–12 weeks, with later stages on the roadmap.
Discovery is the first step in our product strategy capability; user and field research, success metrics, and the roadmap are part of this work. We explain how to narrow the scope proposal coming out of discovery in our article on defining MVP scope.
If you would like to start your project with a discovery sprint, we can assess together in the first conversation whether the scope fits two weeks.