◀ Course contents Part 3 · Module 3-03

Ideation in Design Thinking

Generate wildly, then narrow deliberately

Ideation sits between defining the problem and building a prototype. Done well, it isn't a free-for-all, it's a disciplined swing between opening up as many possibilities as you can (diverge) and narrowing to the few worth testing (converge).

Ready?

1

Ideation's Place in Design Thinking

Ideation is the mode everyone looks forward to, and the one most often run at the wrong moment. Its value depends almost entirely on what happened before it.

In design thinking, ideation sits third — after empathise and define. That order is not ceremonial. Ideas are answers, and an answer is only as good as the question it was generated against.

Deciding where to go on holiday works the same way. Skip that and you have a list of places and no way to choose between them.

The failure this produces is subtle, because a badly framed session still feels productive. Ideas get generated, people enjoy it, a board fills up. What is missing is variety: if the frame was a feature area rather than a user problem, every idea will be a variation on that feature, and the group will never notice because the ideas genuinely are different from each other.

So the question to ask before any ideation session is which point of view it is answering — and if the answer is the name of a screen, the session will produce screens.

1-2

Empathize & Define

Understand users, then frame a sharp problem statement.

3

Ideate

Diverge widely on solutions to that problem, then converge on a few worth trying.

4-5

Prototype & Test

Build a rough version of the strongest idea and check it with real users.

Skip straight from Define to Prototype and you anchor on the first idea instead of the best one.

Where ideation sits Three steps showing ideation in sequence: empathise produces observations, define produces a point of view, and only then does ideation produce answers to that point of view. 1 Empathise observations 2 Define a point of view 3 Ideate answers to that view
Panel three inherits whatever panel two produced. Swap “a point of view” for the name of a screen and the third panel still fills up — same energy, same number of sticky notes, every one of them a variation on that screen. A badly framed session is indistinguishable from a good one while it is running.

Everyday example

Deciding where to go on holiday. Ideation is the twenty minutes of throwing out destinations; it only works once someone has said what the trip is actually for — rest, or seeing family, or somewhere the children will not be bored. Ideas without a frame are just a list of places.

Ideating before defining

A team ran a workshop that produced sixty ideas and shipped two of them. Neither moved anything, because the workshop had opened with "how do we improve the dashboard" rather than a defined user problem. The ideas were creative and the frame was a feature, so every idea was a variation of the same feature.

Quick check

Why not skip ideation and prototype the first solution that comes to mind?

2

The Rules of Divergent Thinking

Divergent thinking is the deliberate act of generating many options before judging any of them. It has rules, and the rules exist because the default behaviour of any group is to converge immediately.

Defer judgement. No evaluation during generation — not even approving noises, which are evaluation wearing a friendly face. Go for quantity. Ask for thirty ideas rather than five good ones, because volume is what gets a group past the obvious. Encourage wild ideas, which move the boundary of what the group considers thinkable. Build on others’ ideas rather than competing with them. Stay on topic, so the session answers the question it was called for.

A generation session runs on exactly the same principle.

The reason these rules need stating is that the failure is social rather than creative. People in a room do not stop having ideas; they stop saying the ones that might be judged. And the ideas most worth having — the odd, half-formed ones — are precisely the ones most exposed to judgement.

What gets lost is invisible, which is why a badly run session feels fine to everyone in it.

Rule 1

Defer judgment

No critiquing ideas during divergence, evaluation happens later, in a separate step.

Rule 2

Go for quantity

More ideas raise the odds a great one is in there. Quality gets sorted afterward.

Rule 3

Build on ideas

"Yes, and", extend a teammate's idea rather than starting over or shutting it down.

Rule 4

Encourage wild ideas

An impossible idea can seed a feasible one. Welcome it before judging it.

Diverge, then converge Two phases shown as four steps: set the rules, generate widely without judging, cluster the output into themes, and only then select. Judgement is withheld until the fourth step. Set the rules no judging, build on others Generate quantity first Cluster find the themes Select now judge
The rule that does the work is set in step one and paid for in step two: no judging while the ideas are still arriving. Nothing is evaluated until the fourth box — because the moment evaluation enters a generation session, people stop offering the ideas most worth having, which are the odd ones.

Everyday example

In improvisation, the rule is "yes, and" — you accept what your partner offered and add to it. The moment someone says "no, but", the scene stops. A generation session runs on the same rule for the same reason.

The senior voice in the room

A session produced almost nothing until the team switched to writing ideas silently on cards before anyone spoke. In the previous format the most senior person had offered an idea first, and the following thirty minutes had been variations on it. Divergence is a social problem more often than a creative one.

Quick check

During a brainstorm, someone says "that would never work" about a wild idea. What's the ideation-rule violation?

3

A Toolkit of Ideation Techniques

A blank page is the enemy of ideation, and every technique worth knowing is really a device for making the page smaller.

How Might We questions turn a problem into an invitation. Crazy Eights gives everyone eight minutes to sketch eight ideas, which forces past the first two that everybody has. SCAMPER walks an existing thing through prompts — substitute, combine, adapt, modify, put to other uses, eliminate, reverse.

Analogous inspiration asks who else has solved this shape of problem: how do hospitals handle handovers, how do airlines handle overbooking? Reverse brainstorming asks how you could make the experience as bad as possible, then inverts the answers.

That last one is quietly the most reliable, and the reason is worth understanding: criticism is far easier to generate than invention. A group that cannot produce ten good ideas will produce thirty bad ones enthusiastically, and each of those inverts into something usable.

Constraints are not the opposite of creativity. They are the usual mechanism for it, and an unconstrained ideation session is the hardest kind to run.

How Might We

Turns the problem into an invitation. “How might we help a new seller price a listing without guessing?”

Crazy Eights

Eight minutes, eight sketches, one per folded panel. Forces everyone past the first two ideas that everybody already has.

SCAMPER

Walks an existing thing through prompts: substitute, combine, adapt, modify, put to another use, eliminate, reverse.

Analogous inspiration

Asks who else has solved this shape of problem. How do hospitals handle handovers? How do airlines handle overbooking?

Reverse brainstorming

Asks how you could make the experience as bad as possible, then inverts the answers. The one to reach for when a group has gone quiet.

Techniques, and what each is for Five ideation techniques and the job each one does: How Might We reframes the problem, Crazy Eights forces past the first ideas, SCAMPER prompts against an existing thing, analogous inspiration borrows a shape somebody else has already solved, and reverse brainstorming — picked out here — turns criticism into ideas. How Might We reframes the problem itself Crazy Eights forces past the first two ideas SCAMPER prompts against an existing thing Analogous inspiration who else has solved this shape? Reverse brainstorming criticism is easier than invention
Every one of these does the same job — it makes the blank page smaller. The keylined row is the one to reach for when a group has gone quiet, and the note beside it is the whole reason: ask a room to break something and the answers arrive immediately, and each one inverts into an idea.

Everyday example

A cook stuck for dinner ideas uses constraints, not freedom: what is in the fridge, what takes under twenty minutes, what would a child eat. Ideation techniques work the same way — they are prompts that make the blank page smaller.

The technique that unstuck it

A team out of ideas was asked the inverse question: "how could we make this experience as bad as possible?" They filled a board in four minutes, laughing. Reversing each item produced a stronger list than an hour of asking for good ideas, because criticism is easier to generate than invention.

Quick check

A team's brainstorms are always dominated by the two most senior, loudest voices. Which technique directly addresses this?

4

Converging Without Killing Good Ideas

Convergence is the half nobody plans for. A session produces forty ideas, everyone feels good, and the ideas quietly die because there was no agreed way to choose between them.

The rule is that convergence must be a separate act, with criteria decided before the options are ranked. Deciding what matters after you can see the options means choosing criteria that favour the option you already like.

Cluster first — many of the forty will be the same idea in different words. Then score against criteria you have named: desirability, feasibility, viability, fit with strategy. Then decide, and decide out loud, so the ideas that lost are visibly parked rather than silently forgotten.

And converging does not mean picking one. Two or three ideas carried into cheap tests is usually better than one carried into a build, because the point of diverging was to have options — and committing to a single idea at the moment of convergence throws away the reason you generated forty.

Converging without killing the good ones Four convergence steps in order: cluster duplicates, agree criteria before ranking, score against them, then park the losers visibly rather than letting them vanish. 1 Cluster many are the same 2 Agree criteria before you see the options 3 Score against the criteria 4 Park visibly losers named, not buried
The keylined step is the one that gets skipped, and skipping it is invisible: criteria chosen once the options are on the wall will quietly favour the option you already liked. Plain voting has the same failure with better manners — everyone’s second choice wins and nobody is excited about it.

Everyday example

Voting on holiday destinations by show of hands picks the least objectionable, not the best. Everyone's second choice wins and nobody is excited. Convergence needs criteria, not a popularity contest.

Dot voting, done right

Give each person a fixed number of votes to place on the sticky notes they find strongest, done silently, all at once, so early votes don't sway later ones. Then sort survivors by explicit criteria: feasibility, desirability, impact.

Convergence needs a method, same as divergence does. "Whatever feels right" just reintroduces the loudest-voice problem you just avoided.

Dot voting, and what it nearly lost

A dot vote surfaced four safe ideas and buried an odd one with a single dot. The facilitator gave each person one "wildcard" dot to spend on something nobody else had backed, and the odd idea came back with three. It shipped that quarter. The first vote had measured comfort, not quality.

Quick check

A facilitator ends ideation after five minutes because "we already have three good ideas." What's the risk?

Drill what you learned
Scenario 1 easy

In a brainstorm, a manager says "let's only write down ideas we could ship next sprint."

What's the issue?

Scenario 2 easy

A remote team's ideation call is a free-for-all; three people talk the whole time and five stay silent.

What would improve this?

Scenario 3 easy

The team has 40 sticky notes and immediately picks the first 3 anyone reads aloud, to save time.

What's wrong?

Scenario 4 easy

A facilitator introduces Worst Possible Idea and the team is confused why they'd brainstorm bad ideas on purpose.

What's the value of this technique?

Scenario 5 easy

A junior PM keeps saying "yes, but" in response to teammates' ideas during a brainstorm.

What's the fix?

Scenario 6 medium

A team runs SCAMPER on an existing subscription-box product and gets to "E: Eliminate."

What should they ask?

Scenario 7 medium

After a Crazy 8s round, one panel is illegible scribbles the sketcher can't even explain.

What should the team do with it?

Scenario 8 medium

A stakeholder insists ideation should only include ideas technically feasible at the outset.

How does this affect divergence?

Scenario 9 medium

A group converges on the very first idea proposed, five minutes into a planned 45-minute session.

What's the concern?

Scenario 10 medium

A team spends 3 hours generating ideas and refuses to stop, worried they'll miss "the best one."

What's the risk on the other side?

Scenario 11 hard

A dot-voting session ends with three ideas tied: one ambitious, two safe and similar.

How should the team break the tie?

Scenario 12 hard

Someone proposes a wild, seemingly-impossible idea and the room goes quiet.

What should the facilitator do?

Scenario 13 hard

A team never captures ideas anywhere — they just talk, and by the end half the ideas are forgotten.

What's the fix?

Scenario 14 hard

A senior exec joins an ideation session and immediately says which idea they personally prefer.

What's the risk?

Scenario 15 hard

A team converges using a show-of-hands vote, and the two most senior people's votes visibly sway the room before others vote.

What's a better convergence method?

Drilled it. Now apply it to a real situation.

Put it to work
From lesson 1

IDEO redesigned a shopping trolley in five days

IDEO

In 1999 ABC's Nightline gave IDEO five days to redesign the supermarket shopping trolley on camera. The team spent the first day not designing at all: they went to supermarkets, watched shoppers, and talked to store managers about theft and to parents about child seats.

Only after that did anyone sketch. The ideation session that followed produced dozens of concepts, which were cut down to a handful and built as rough prototypes by day four.

Why did a five-day deadline still begin with a day of observation rather than ideas?

From lesson 2

Ninety minutes, four ideas

A retail bank

Eight people, one facilitator, four ideas at the end — three of them variants of the same thing. The transcript shows what happened.

What happened in the room
MinuteEvent
4First idea proposed by the most senior person present
6Second idea met with "we tried that in 2019"
11Discussion of technical feasibility begins
23Cost estimate requested for idea one
38Two people have still not spoken
61Room converges on a variant of idea one

Which rules of divergent thinking were broken?

Select all that apply — there are 3 to find.

From lesson 3

Picking the technique to fit the stuck-ness

A hospital appointment system

Four sessions, four different kinds of stuck. Each needs a different technique — using the same one every time is why most teams think ideation does not work for them.

Match each situation to its technique, in the order the situations are listed.
1. Every idea is a variant of the current booking form. 2. The team has 60 ideas and no way to talk about them. 3. Nobody can see past "patients should just phone earlier". 4. Two strong ideas, and the team is split.

Drag the rows, or use the arrows, then check.

  1. Provocation — "how would we do this if appointments could not be booked in advance at all?" Situation 1. Deliberately removing the thing everyone is anchored on is what breaks a room out of variants. The provocation does not have to be workable; it has to be unignorable.
  2. Affinity clustering — group the 60 on a wall and name the groups Situation 2. Volume without structure is unusable. Clustering turns sixty items into six themes the room can actually reason about, and the theme names often become the real insight.
  3. Reframing with How Might We — "how might we make a late booking as easy as an early one?" Situation 3. When the room blames the user, the frame is wrong. A How Might We turns the complaint into a design question and moves the burden back to the product.
  4. Prototype both cheaply and let the evidence decide Situation 4. A split between two strong ideas is not an ideation problem at all. More discussion will not resolve it; a day of testing will.
From lesson 4

Converging without killing the good ones

A children's reading app

Forty-one ideas from a strong divergent session. Three hours to get to a shortlist. Here is how the room proposes to do it.

Work through the convergence.

  1. Step 1 of 3

    First proposal: everyone votes, top five survive. What goes wrong?

    So the room needs a way to keep the unfamiliar ideas alive long enough to be understood.

  2. Step 2 of 3

    Second proposal: cluster the 41, then vote within each cluster. Better?

    Five clusters, one survivor each. Now the shortlist has to be judged against something.

  3. Step 3 of 3

    What should the five be judged on?

    Two of the five failed the "works for a non-reading parent" criterion, which nobody had noticed while the ideas were being admired.

Notification