Crazy 8s brainstorming: the 8-minute sprint, explained

Crazy 8s brainstorming is a design sprint exercise where each person sketches 8 distinct ideas in 8 minutes, one per minute, on a sheet folded into 8 panels. The timebox is the point: 60 seconds per idea leaves no room for polishing, so you get range instead of one refined favorite.

The exercise comes from the Google Ventures design sprint, where it sits in the middle of sketching day. But it works outside sprints too: for landing pages, feature concepts, email subject lines, even naming. Here is how to run it properly, where it usually breaks, and a variant for teams who freeze at the word "sketch."

What Crazy 8s is actually for

Crazy 8s is a forced-divergence tool. Most people, given a design problem, produce one idea and then spend the session decorating it. The folded sheet makes that impossible: eight empty panels sit there demanding eight different answers, and the one-minute cadence means your inner critic never gets the floor.

Two things it is not. It is not a drawing contest; stick figures and boxes are the expected fidelity, and anyone who produces something beautiful in 60 seconds was probably reusing an old idea. And it is not a full session; Crazy 8s generates raw material in 8 minutes, and you still need a converge step afterward. On its own it fits the diverge half of a session; the full arc of prep, rules, and timing is covered in how to run a brainstorming session. The exercise came out of design sprints and still lives mostly with design teams, which is the audience our page on brainstorming for designers is written for.

What is a Crazy 8 sketch?

A Crazy 8 sketch is one of the eight rough drawings produced in the exercise: a single panel, drawn in about 60 seconds, showing one distinct idea. It is a thumbnail, not a design. Boxes, arrows, stick figures and three words of label are the expected fidelity, and the whole point is that it takes less than a minute to make and less than ten seconds to read.

The confusion worth clearing up is that a Crazy 8 sketch is not a wireframe or a screen mockup. A wireframe commits to layout and asks to be critiqued on it. A Crazy 8 panel commits to a concept and deliberately leaves the layout undecided, because deciding it costs time you do not have and invites feedback about button placement when you are still choosing a direction. If your panels look polished, the timer is too long or you were drawing something you had already designed in your head.

Are there variants with a different number of ideas?

Yes, and the number is less sacred than it looks. The standard is eight panels in eight minutes, one minute each. Compressed versions run the same fold at 30 or 45 seconds a panel, which suits a remote session where attention is shorter. Extended versions push to 15 or 16 ideas across two rounds with a short break between, which is worth doing when the first eight all came out looking like variations on the same theme.

The two-round version is the one most teams should know about, because the second round is where the useful material usually is. The first eight panels drain the obvious answers that everybody already had. After a two minute break, the second round has to go somewhere else, and that is when the genuinely different directions appear. What does not work is simply extending the timer: giving people three minutes a panel produces eight prettier versions of the same eight ideas, not better ones. The pressure is doing the work, not the clock.

Setup: 5 minutes before the timer starts

1. Frame one specific challenge. "The pricing page for our fitness app" works. "Our app" does not. Everyone sketches against the same prompt.
2. Fold the paper. One sheet per person, folded in half three times: eight panels. The fold matters more than it looks; eight visible empty boxes apply pressure a blank page never does.
3. Markers, not pens. Thick markers force low fidelity, which is what you want. Fine pens invite detail, and detail eats the minute.
4. State the rules out loud. Eight different ideas, not eight versions of one. Ugly is fine. No talking during the eight minutes. Nobody's sheet is shown until the end.
5. Optional: 3 minutes of silent note review first. In a full sprint, Crazy 8s follows research and lightning demos. Standalone, give people 3 minutes with the brief so the first minute is not spent understanding the prompt.

Timing: how the 8 minutes run

Set one timer for 8 minutes, or call out each minute. Calling minutes is better with groups new to the exercise; the "next panel" prompt stops people from camping on a favorite.

The struggle in the middle is the exercise working. The first two panels are memory; panels six through eight are invention. If everyone finishes comfortably at minute six, your prompt was too narrow or your team sketched variations, not ideas.

Panels one and two are memory. Panels six through eight are invention.

After the timer: converging the sheets

Do not go around the table explaining sheets; presenting invites pitching, and pitching rewards charisma over ideas. Instead, tape every sheet to the wall, give the group silent minutes to read, and have each person place two or three dot votes on individual panels, not whole sheets. Then discuss only the top-voted panels.

A worked example: an agency running Crazy 8s on names for a client's carbon-accounting product got 40 panels from five people. Dot voting surfaced six panels, four of which were from minutes six to eight of somebody's sheet. The winning direction, a ledger metaphor, appeared independently on two sheets, which the team read correctly as a signal, not a coincidence. Clustering repeats like that before voting is exactly what an idea clustering pass is for when the panel count gets large.

Failure modes, and the fix for each

• Eight variations of one idea. The most common failure. Fix: say it in the rules, then enforce it in minute three: "if panel four looks like panel three, cross one out."
• The art contest. One strong sketcher makes everyone else self-conscious. Fix: thick markers, and show an intentionally terrible example sheet before starting.
• Stalling at panel five. Fix: keep a list of prompt flips ready to read aloud: "what would the mobile-only version be," "what if it had no text," "what would the expensive version look like."
• Talking during the round. One comment resets everyone's thinking. Fix: silence is a stated rule, and the facilitator holds it.
• Skipping the converge. Forty panels photographed and forgotten is the quiet failure nobody logs. Fix: book the voting into the same meeting, before anything else happens.
• Vague prompt. Eight minutes is too short to also figure out the question. Fix: write the challenge sentence on the wall before folding paper.

How do you do Crazy 8s without sketching?

Plenty of good problems are not visual. A SaaS team attacking month-two churn, a newsletter hunting for growth loops, a founder naming a company: none of these want stick figures. The words variant keeps the mechanism and drops the sketching.

Same fold, same 8 minutes, but each panel gets a written idea in a fixed format: one headline line plus one mechanism line. For the churn prompt, a panel might read "Day-10 rescue email / triggered when a user has not logged in for 5 days, shows the report they have not seen." The two-line format is the discipline; a headline alone is a slogan, and the mechanism line forces the idea to exist.

Everything else stays identical: silence, minute calls, wall, dots. The variant also works async and remote, where sketch quality varies wildly but two lines of text are democratic. If your team is distributed, running the round in an online brainstorming tool keeps the timebox honest and gives everyone the same written starting material instead of eight webcam photos. Brainstormer will not run your dot-vote for you, but it will hand the room around twenty tagged directions and the three strongest with the reasoning attached, so the debate starts from arguments rather than from a blank fold.

Solo Crazy 8s

The exercise does not need a room. Solo, the fold and the timer do the same work: eight panels, eight minutes, no self-editing. The failure mode solo is softer deadlines, so use a real timer and treat the bell as law. Solo rounds pair well with a second pass an hour later; your panels one and two will be different the second time, because the first round drained the obvious layer. For more fast-divergence patterns that stack with this one, see the wider toolkit in rapid ideation.

The short version

1. One specific challenge, written where everyone can see it.
2. One sheet each, folded to 8 panels. Markers.
3. 8 minutes, one idea per minute, silence, no variations.
4. Wall, silent reading, dot votes on panels.
5. Discuss only the winners. Decide something before the meeting ends.

Eight minutes of forced range beats an hour of polishing the first idea. That trade is the whole exercise.