Run a Design Sprint

A design sprint is a five-day workshop: one product question, a prototype, and a test with real people by Friday. Use this when talking will not settle the bet and you need a cheap yes or no before a quarter of build.

Use this

  • The team is stuck between two or three directions and another meeting will not settle it
  • You need a cheap no before a quarter of engineering time
  • A decision-maker can give you one focused week, then leave

Skip this

  • You still do not know who the user is. Run interviews first
  • The question is bigger than five days, like whether to enter a market
  • Nobody with authority can sit in on Wednesday
  • You already have a flow and just need a usability test

Steps

Six steps, one job per day. Planning and the challenge sit together; Friday’s test and shareout sit together.

01Set the challenge

Write one problem the team can answer in five days, then lock the roster, the decision-maker, and Friday’s testers. A sprint without testers booked is a workshop.

Do this 1–2 weeks ahead. The decision-maker has to be in the room on Day 3.

02Map the problem

03Sketch solutions

04Decide and storyboard

The decision-maker picks one concept and a prototype scope. Then storyboard the critical path: 5–7 scenes, happy path plus one error. Be explicit about what you are not building.

Do not prototype without this person in the room.

05Build the prototype

06Test and share

Frequently asked questions

We only have three days. Do we still run this?

Keep the spine: problem, one concept, a prototype, and a test. Cut competitive review if you already know the space. Do not cut testers.

We do not have existing research. Can we still sprint?

Do three to five interviews the week before, or start from support tickets and analytics. A sprint uses evidence you already have. It is not a research study.

Does this work remotely?

Yes. Same prompts. Shorter blocks, one person driving the board, cameras on for decide and test.

What happens after Friday?

If the test worked, write the PRD. If it did not, you still got a cheaper no than building the thing.