I've run a lot of product sprints. Most of the ones that flopped didn't flop on ideas — they flopped on setup. Here's the checklist I lean on to make a sprint actually produce something.

Before the sprint

The real work happens before anyone walks in the room.

  • Frame a sharp challenge: "improve onboarding" is useless. "Cut signup drop-off by 30%" is a target. Write it as a question: "How might we cut signup drop-off by 30%?"
  • Get the right five people: a decider who can actually commit, a designer, an engineer, someone who knows the customer. Five is the number. Past seven, it stalls.
  • Be honest about the output: a sprint doesn't ship production code. It produces a tested prototype and a real decision. Say so up front.
  • Block the calendar and defend it: five days, same hours. Sprints die when people treat them as optional.

During the sprint

Keep the rhythm tight:

  • Day 1 — Understand: map the problem, talk to stakeholders, set the metric. End knowing what you'll prototype.
  • Day 2 — Diverge: generate, sketch, go wide. Don't converge yet.
  • Day 3 — Decide: critique, pick one path, storyboard it. This is where teams stall. Push through.
  • Day 4 — Prototype: build something testable. It doesn't need to be pretty, just real enough to get an honest reaction.
  • Day 5 — Test: five users, one at a time. Watch. By the third, the pattern is usually obvious.

After the sprint

This is where most of the value leaks out. Don't let it:

  • Write the decisions down within 24 hours: what you learned, what you'll build, what you're killing.
  • Book the follow-up now. A sprint creates momentum; protect it with a two-week check.
  • Share it loudly: the people who ran it know the most, so make them tell leadership, engineering, support. Don't bury it in a deck.
Run a sprint when you're stuck between options or alignment is broken. Don't run one when you already know the answer.

A good sprint doesn't only hand you a prototype. It hands you clarity. That's the part that lasts.