An HR team has half an hour to make a new hire’s first week less confusing. The most senior person suggests a welcome video, everyone nods, and the rest of the meeting goes into polishing that one idea. Nobody writes anything alone, and nobody gets past the first answer in the room.

That meeting skips most of what research suggests about how to come up with ideas: produce many before judging any, write alone before talking, change your state when you stall, force unusual connections, and get help choosing. Few techniques have been tested against each other, and most tests were short lab sessions, so each one below carries an evidence label. Here are twelve, roughly from best tested to least:

1. Write ideas alone before the group talks
2. Write in parallel, then read each other’s lists
3. Keep going after the obvious ideas run out
4. Take the problem for a walk
5. Step away after a real attempt
6. Borrow a solution from a distant field
7. Combine two things that don’t belong together
8. Run a prompt list over an existing idea
9. Add one constraint on purpose
10. Map ideas outward from a single word
11. Ask an AI for a first batch, then move away from it
12. Let peers help you pick

Creativity training works on average, and in a 2004 meta-analysis by Ginamarie Scott, Lyle Leritz and Michael Mumford, programs that taught thinking skills, such as generating and combining ideas, did better than most other approaches.¹

What the studies behind idea techniques measure

Most research on idea techniques measures one narrow thing: how many ideas people list in a few minutes, and how unusual those ideas are. That is only the idea-generation part of creativity, one of several ways psychologists define and test creative thinking. A technique can raise that score without helping anyone finish a real project. A workshop that ends with 60 sticky notes has done well on the thing the studies count; whether one of those notes becomes a better onboarding process is a separate question.

So each technique below carries an honest label. “Trial” means people were randomly assigned to use it or not; “expert opinion” means the technique is widely taught but nobody has tested it directly for producing ideas. The best overall evidence that techniques can be learned comes from reviews of creativity training.

The study

Moderate evidence

Can idea skills be trained? A 2004 pooling of 70 studies

The authors pooled 70 studies of creativity training programs. Well-designed training generally improved performance, and the gains held across different measures, settings and groups of people. Programs that taught cognitive skills, such as finding problems, generating ideas and combining concepts, and that gave people realistic practice, tended to do best.¹

The practical reading is encouraging: idea generation behaves like a skill, and the techniques that work share a shape. The caveat is size. A 2024 review of five decades of training studies by Ut Na Sio and Hugues Lortie-Forgues also found gains on average, but warned that many studies were weak and that published resultspublication bias: The tendency for studies with striking or statistically significant results to be published, while quieter ones stay in the drawer. Since the missing studies are mostly the unimpressive ones, a pooled result tends to overstate the effect until it is corrected.Full entry in the glossary may overstate the benefit.² Expect a nudge, not a transformation.

Myth

Some people are simply not idea people, and no method will change that.

Fact

Idea generation responds to practice: a 2004 meta-analysis pooling training studies found gains across settings and groups of people.

How to come up with ideas when the first ones run dry

The three best-tested techniques all protect one thing: time to generate before anyone judges. A 2010 University of Pennsylvania experiment, a 2000 study of written group work and lab studies of long idea lists point the same way: get many ideas down, independently, before anyone evaluates them.

Write ideas alone before the group talks

If a team needs ideas, give everyone a few minutes of silent, solo writing before any discussion. In the Pennsylvania experiment, students who worked alone first and then pooled their ideas as a group came up with more ideas, and better ones, than teams who worked together from the start.³ The reasons spoken sessions lose ideas are laid out in the research on group brainstorming. The lesson is simple: nobody should hear an idea until everyone has written their own.

Write in parallel, then read each other’s lists

Once people have written alone, let them read the others’ lists and add to them, still in writing. In a 2000 experiment by Paul Paulus and Huei-Chuan Yang, groups who wrote ideas on slips and passed them around produced more ideas than an equal number of people who worked separately and pooled their lists.⁴ Reading someone else’s idea can remind you of a category you had not considered, without the wait for a turn to speak. On a remote team, a shared document where everyone types at once may do a similar job.

Keep going after the obvious ideas run out

Your first ideas are usually everyone’s first ideas. In lab studies by Roger Beaty and Paul Silvia, ideas tended to get more original the longer people kept at a task, a pattern psychologists call the serial order effect.⁵ The early answers come from easy, well-worn associations; the later ones take effort. Ask a team how to cut its email load and the first answers are the ones any team would give: fewer cc’s, a no-email afternoon. Ideas that fit your particular team usually take longer to surface. When the list slows down and feels empty, that is often the start of the better part, not the end.

Changing your state: a walk or a break

When you are stuck, changing what your body is doing can help, and it costs little to try. Two lines of evidence support this: 2014 Stanford experiments on walking, and a 2009 meta-analysis of incubation studies showing that breaks help on average, more on open-ended tasks and less when the break is filled with demanding work.

Take the problem for a walk

In four experiments, Marily Oppezzo and Daniel Schwartz asked adults to list unusual uses for everyday objects while sitting or while walking, on an indoor treadmill or outdoors. About four in five people scored higher on creativity while walking than while sitting, and the boost lingered briefly after they sat back down.⁶ Walking helped far fewer people on puzzles with one right answer, so use it to generate options, not to check a calculation. A walking one-to-one, or pacing during a call while you list options, is the everyday version.

Step away after a real attempt

A 2009 meta-analysis of incubation studies by Ut Na Sio with Thomas Ormerod reported that breaks helped on average. The benefit was largest on open-ended tasks, grew with a longer first attempt, and shrank when demanding work filled the break.⁷ The conditions matter: work on the problem properly first, then switch to light tasks such as tidying or a routine errand. Our guide to the incubation effect covers when a break helps most.

Making unusual connections

Five popular techniques try to jolt you out of familiar associations by importing something from outside the problem: another field, an unrelated object, a checklist of prompts, a limit or a visual map. Their evidence ranges from classic lab experiments on analogies to almost no direct testing, so treat the last three as useful structure.

Borrow a solution from a distant field

Ask: who else has solved a problem shaped like mine? In classic 1980 experiments, Mary Gick and Keith Holyoak gave students a story whose solution matched a later medical puzzle. Many used the story only once they were told it might help.⁸ Analogies work, but people rarely spot them unprompted. So make the search deliberate: a team speeding up hospital handovers might study how racing crews change tires, then ask what made that fast.

Combine two things that don’t belong together

Pick two unrelated ideas, products or features and ask what a mix of them would do. Combining concepts was one of the thinking skills taught in the most effective programs in the Scott, Leritz and Mumford meta-analysis.¹ That supports it as part of training packages; nobody has tested it alone as a technique. A bakery that crosses its loyalty card with a book club, for example, has two ordinary things and one new offer.

Run a prompt list over an existing idea

Prompt lists such as SCAMPER (substitute, combine, adapt, modify, put to another use, eliminate, reverse) ask a fixed set of questions of something that already exists. SCAMPER itself appears never to have been put through a controlled trial for idea generation, so its evidence label is expert opinion. Its value is practical: it stops a session from running out of angles. Use it on a product, a process or a meeting format, and cross out prompts that plainly don’t apply.

Add one constraint on purpose

Limits cut both ways, depending on the kind. A 2019 review of constraint research across management, marketing and other fields by Oguz Ali Acar, Murat Tarakci and Daan van Knippenberg found the findings fragmented and conflicting, with different constraints working through different mechanisms.⁹ A deadline, a tight budget and a rule need not act alike, so read the evidence as a prompt to experiment, not a recipe. Try one limit at a time: “it must cost nothing”, “it must fit on one page”, “no new software”.

Map ideas outward from a single word

A mind map puts one word in the center and branches associations outward. The evidence for maps comes mainly from studying: a 2006 meta-analysis by John Nesbit and Olusola Adesope linked learning with concept maps to better retention, with effects from small to large depending on how the maps were used.¹⁰ Whether mapping produces more or better ideas than a plain list has not been settled. Use it if seeing branches helps you, and switch to a list if it doesn’t.

1. Go wide: alone and in writing first, and keep going after the obvious ideas run out
2. Step away: a walk or a light task, after a real attempt at the problem
3. Choose: ask peers who do similar work to judge, not only yourself or the boss

Using AI and other people’s judgment

The last two techniques bring in outside minds: an AI assistant to widen the first batch, and colleagues to help pick. A 2024 online experiment on AI story ideas and a 2016 study of circus professionals each point to the same rule: get outside help, but keep the judgment where it is most accurate.

Ask an AI for a first batch, then move away from it

Generative AIgenerative AI: AI systems that produce new text, images, audio or code in response to a request, by generating output that resembles the data they were trained on. Chatbots built on large language models are the best-known kind.Full entry in the glossary can produce a long first list in seconds. In a 2024 online experiment by Anil Doshi and Oliver Hauser, writers randomly assigned to receive GPT-4 story ideas wrote stories that readers rated more creative, with the biggest gains for the least creative writers. The AI-aided stories were also more similar to one another.¹¹ If many people ask the same tool, they get similar ideas. Treat the list as raw material: cross out what you’d expect a competitor to get, and push on what’s left. For the broader picture, see how AI assistants help and fall short at work.

Let peers help you pick

Choosing is its own skill, and you may not be your own best judge. In a 2016 field study of circus professionals by Justin Berg, forecasts of how new acts would do were checked against audience data. Creators predicted the success of other creators’ novel acts more accurately than managers did, but held no such edge for their own.¹² The study covered one field, so treat it as a lead. Before you commit, ask one or two peers who do similar work which ideas they’d back.

All twelve side by side, from firmest evidence to thinnest

The techniques with the firmest support separate generating from judging and change how you generate: writing alone first, pushing past the obvious, walking and taking breaks. Distant analogies and combining concepts have support in experiments and training reviews; the popular prompt tools rest mainly on expert opinion.

Three habits sum up the table: separate producing ideas from judging them, change your state when you stall, and borrow judgment from people who do similar work.

WiserHours

An idea session in six steps

• Write the problem as one sentence
• Write ideas alone, in silence, before any discussion
• Keep going after the list feels empty
• Walk or take a light break, then add more
• Try one distant field, one combination and one constraint
• Ask one or two peers to pick their favorites

Find this checklist online:
https://wiserhours.com/creativity/how-to-come-up-with-ideas/#checklist-how-to-come-up-with-ideas-session

The bottom line

Good ideas come less from inspiration than from an order of work: generate many before you judge any, alone and in writing first, and keep going past the obvious ones. When you stall, walk or take a light break rather than forcing it. Use analogies, combinations, prompt lists and AI to widen the list, and let people who do similar work help you choose.