In this guide
- 01Prepare the decision
- 02Explore and choose
- 03Prototype and rehearse
- 04Observe and interpret
A design sprint is a focused working process that turns one important question into a prototype and a structured test with relevant users. Prepare it around a decision the team can genuinely make. A busy week of sketching is useful only if the resulting evidence can change what happens next.
Choose a sprint when the question is sufficiently bounded, the necessary people can participate and a prototype can expose an important uncertainty. Use discovery research first when the team still needs to understand whose problem matters or what situation it should investigate.
Distinguish a sprint from a compressed project
GV’s design sprint explanation describes a five-day route through framing, sketching, deciding, prototyping and user testing. Google’s Design Sprint Kit names six methodological phases: Understand, Define, Sketch, Decide, Prototype and Validate. These are related accounts, not contradictory claims that every sprint must contain the same number of calendar days.
Be explicit about the format you are using and why. A short workshop using sprint methods may be valuable, but it should not imply that research, a tested prototype and a production decision all happened if they did not.
A sprint does not replace engineering, accessibility work, operational preparation or a business case. Its strength is a focused learning commitment. If the team is already required to build a fixed solution regardless of what users do, discuss that constraint before scheduling the sprint.
Prepare the decision and the test
Consider a hypothetical museum investigating whether visitors can choose an appropriate ticket without assistance. The team suspects that combining entry time, exhibition access and concession eligibility is confusing. A sponsor can authorize further design work, but cannot change ticket policy during the sprint.

Write the question narrowly enough to test: can intended visitors understand the proposed choices and select a ticket for a realistic visit? Record what the sprint will not answer, such as future attendance, operational reliability or the affordability of a pricing policy.
Prepare the team before the first session. Include knowledge of visitors, service operations, design and the relevant technical or policy constraints. Name the person who can decide which concept to prototype. If that person cannot attend the decision point, arrange a clear delegation or choose another time.
Recruit users before the prototype exists. Define the situations the test needs to cover and arrange accessible participation. A prototype that is ready on Thursday does not solve the problem of having nobody relevant available to try it on Friday.
Agree the boundaries for participant information, recording and consent. The sprint’s time pressure does not reduce the need to treat people and their data carefully. The customer-interviews guide explains how to separate a research conversation from a sales pitch.
Organize the week around useful outputs
An original planning outline for the museum might be:
| Working period | Main task | Output needed before moving on |
|---|---|---|
| Frame | Review existing evidence and the visitor journey | One testable question and explicit constraints |
| Explore | Develop different responses independently | Distinct, understandable concept sketches |
| Decide | Examine trade-offs with the authorized decision maker | A chosen test scope and storyboard |
| Prototype | Build enough of the experience to attempt the task | A rehearsed prototype and research script |
| Test and interpret | Observe use, compare accounts and decide next work | A bounded finding record and a new commitment |
This outline is a practical adaptation, not a reproduction of a proprietary sprint worksheet. Preserve the learning sequence when adapting the schedule. Shortening the calendar is reasonable only if the required preparation and outputs still fit.
Use brainwriting or another suitable method to prevent the first confident sketch from becoming the only option. Ask people to explain how their concept addresses the same question. Do not compare an attractive visual concept with an incomplete alternative and call the choice decisive.
Build only what the question needs
GOV.UK’s prototyping guidance distinguishes exploratory prototypes from production services. The team should choose the form that allows the intended question to be investigated. A paper sequence may expose ticket-language problems; a clickable prototype may be needed to examine navigation between choices.
For the fictional museum, show the relevant ticket information, the selection steps and a believable consequence of the choice. Avoid spending the prototype period perfecting screens that the test will never use. Rehearse the whole task with a colleague to find broken links, missing states and facilitator instructions that accidentally reveal the answer.
Keep the prototype clearly separate from a live service. Participants should not believe they have purchased a real ticket unless that is deliberately part of a properly prepared operational test. A realistic-looking surface is not evidence that payment, capacity or support systems work behind it.
The MVP-versus-prototype comparison helps distinguish a learning artifact from a usable first offer. The distinction matters when a sponsor asks why the prototype cannot simply be launched after the sprint.
Use the result without overstating it
During the test, give participants a realistic task and observe before explaining. GOV.UK’s moderated usability testing guidance supports investigating how people attempt tasks and where they encounter problems. Enthusiasm about the visual design is different evidence from successfully choosing an appropriate ticket.

Separate observations, interpretations and recommendations. Record that a participant returned to an earlier step, then investigate why; do not immediately conclude that a particular label caused the difficulty. Note facilitator help, prototype limitations and situations the session did not cover.
Use a simple results record:
| Field | Example question |
|---|---|
| Task | What was the visitor trying to arrange? |
| Observation | What did they actually do or say? |
| Interpretation | What explanation fits, and what else could explain it? |
| Limitation | What did the prototype or participant coverage leave unresolved? |
| Decision | Revise, investigate another question, proceed to a larger commitment or stop? |
No outcomes are asserted in this hypothetical example. A small qualitative test can reveal problems worth addressing; it does not estimate how often every visitor will experience them. Use the sampling guide if frequency is the next decision question.
Common failure modes
A sprint can become an approval ceremony when the preferred solution is chosen beforehand. It can become a design performance when the prototype is polished but the test question remains vague. It can also leave a convincing presentation with no owner for the next inquiry.
Prevent these failures through explicit decision conditions. Before testing, ask what observation would justify a revision or a stop. After testing, identify the strongest uncertainty that remains. Use pivot-or-persevere guidance when interpreting mixed evidence rather than treating every result as permission to build.
Evidence limits and follow-through
The originators’ accounts explain the method; they do not establish a universal success rate or prove that a five-day format is optimal. The proposed museum sprint is an original fictional example. Its schedule and prompts are planning aids, not a validated intervention.
End with a decision that names the evidence, remaining uncertainty, responsible person and next commitment. If the team needs further discovery, record it plainly. A useful sprint can reveal that a problem needs a different question, provided someone acts on that finding.
Sources:
- GV (n.d.). The Design Sprint.
- Google (n.d.). Design Sprint Methodology.
- Government Digital Service (n.d.). Using moderated usability testing.
- Government Digital Service (n.d.). Making prototypes.



