Lightning Decision Jam: the seven steps, timings and rules

A Lightning Decision Jam is a structured workshop that takes a team from problems to prioritized action items in about an hour, with almost no open discussion. Participants write silently, vote with dots, reframe the winning problem as a How Might We question, generate solutions, then plot those solutions on an effort and impact grid. Talking is limited on purpose, because open debate is what makes normal meetings run long and end undecided.

The format was popularized by the design agency AJ&Smart, who built it out of design sprint methods for teams that could not spare five days. It spread because it fixes the specific meeting everyone hates: the one where a problem gets aired for fifty minutes and the outcome is another meeting. This article walks the steps, the timings, what each rule is doing, and where the format breaks.

What is a Lightning Decision Jam?

A Lightning Decision Jam, usually shortened to LDJ, is a facilitated session for solving a team problem quickly using silent writing, dot voting and a prioritization grid. It runs 30 to 60 minutes for a small team, needs one facilitator who does not contribute ideas, and produces a short list of action items with owners rather than a decision document.

The design principle is simple: replace discussion with parallel work wherever possible. In a normal meeting one person talks and everyone else waits, so a team of six gets one person's output per minute. When everyone writes at once you get six. The facilitator's job is to protect that structure, not to steer the content, which is the same discipline behind a well-run innovation workshop.

What are the steps of a Lightning Decision Jam?

Seven steps, run in order, each time-boxed. The sequence matters more than the exact minutes: every step narrows what the room is working on, so skipping one leaves the next step with too much input.

Two closing steps are worth adding even though they are not always listed. Convert whatever lands in the high-impact, low-effort corner into action items with a named owner and a date, and agree how you will know in two weeks whether it worked. Without those, an LDJ generates a satisfying hour and no change.

Why is discussion banned during the session?

Because discussion is where the time goes and where the bias enters. Open debate in a group rewards whoever speaks first, speaks most confidently, or outranks everyone else, and the research on group brainstorming has shown for decades that this suppresses the total number of ideas. Writing in parallel removes the queue, and dot voting removes the need to argue a preference out loud.

The ban is not permanent. Steps 2 and 7 allow short clarifying exchanges, and the action-item conversation at the end is deliberately open. What is banned is evaluating an idea at the moment it appears, which is the habit that quietly kills range. We wrote about the underlying mechanism in why group brainstorming fails, and it is the same reason brainwriting outperforms a round-table.

How many people should be in a Lightning Decision Jam?

Four to eight participants plus a facilitator. Below four you lose the voting signal, since two dots each does not separate anything meaningful. Above eight the writing steps stay fine but presentation and voting stretch past their time boxes and the session loses its pace. If you need more voices, run two parallel jams on the same problem and merge the solution sets before step 6.

Cross-functional beats homogeneous here. A room of engineers will generate engineering solutions to what is often a communication problem. Including someone from support or sales tends to change which problems win step 3, and that is usually the most valuable thing that happens all hour.

Can you run a Lightning Decision Jam remotely?

Yes, and remote is arguably the better setting. The format is already silent-first, so it loses nothing over video, and a shared board removes the handwriting problem. Two adjustments help. Extend each writing step by a minute or two, because typing into a tool has more friction than grabbing a sticky note. And make voting genuinely anonymous, since visible dots on a shared board reintroduce exactly the follow-the-leader effect the format is built to prevent.

Distributed teams often go one step further and run steps 1 and 5 asynchronously in the day before the call, then use the live hour purely for voting, the grid and the action items. That variant halves the meeting and generally improves the input, since people think about the problem across a day instead of seven minutes. Our guide to online brainstorming covers the async setup, and the meeting agenda template gives you a script to paste into the invite.

Lightning Decision Jam versus a design sprint

A design sprint is four or five days, involves prototyping and user testing, and is meant for a substantial bet where being wrong is expensive. An LDJ is an hour, produces action items rather than a tested prototype, and is meant for a recurring team problem. They are not competitors: the jam is what you run when the sprint is not justified, which is most weeks.

Use the jam when the problem is internal and the solutions are things your team can act on directly. Use a sprint when the question is whether customers want something, because no amount of internal voting answers that. And if the recurring problems your jams keep surfacing are structural rather than tactical, month after month, the honest read is that the format is the wrong instrument and you need a proper look at how the organization's processes are actually performing.

What goes wrong in a Lightning Decision Jam?

Four things, in rough order of frequency. The facilitator participates, which turns the session into a guided tour of their preferred outcome. Fix it by having someone facilitate who has no stake, or by rotating the role.

The time boxes get treated as suggestions. Once one step runs long the whole session drifts and the grid gets rushed, which is the step that actually produces the decision. Run the clock visibly and cut steps off mid-sentence: it feels rude the first time and works every time after.

The How Might We question comes out too broad. "How might we improve onboarding" produces vague solutions, while "how might we get a new user to their first import in under five minutes" produces specific ones. Getting that scope right is a real skill, and we broke it down in how to write a How Might We question.

And the solution step produces obvious answers. Seven minutes of silent writing gets you what the room already knew, which is fine for tactical problems and thin for anything harder. When step 5 keeps returning the same four ideas, the constraint is not the format, it is the input.

Widening step 5 when the room is out of ideas

The generation step is the one place an LDJ has no mechanism. Steps 3, 6 and 7 are all narrowing, so if step 5 produces a weak set the session efficiently selects the best of a weak set. Facilitators usually notice this as the moment the energy drops and people start writing variations on each other's notes.

The fix is to bring structured range into that seven minutes. Feed the How Might We question from step 4 into Brainstormer and it returns dozens of tagged directions, each from a forced angle: invert the problem, remove a step, borrow a mechanic from another industry. Put those on the board alongside what the team wrote, and step 6 votes on a genuinely wider field. The grid at step 7 then works the way an impact effort matrix is supposed to, sorting real alternatives instead of near-duplicates. Try your own How Might We question in the studio above before the next session.