When a brainstorm stalls, the problem is rarely a lack of ability — it is that the mind keeps circling the same familiar paths. An Object Generator breaks that loop by forcing an unexpected input into the process. Designers, writers, inventors and problem-solvers have used this technique for decades, sometimes called random input or forced association.
This guide explains why an unrelated object unsticks thinking, gives you a repeatable session structure, and covers the mistakes that make the technique feel like a gimmick rather than a tool.
Why an Unrelated Object Helps
Thinking follows worn grooves. Ask a team to improve a coffee shop and you will get faster service, better coffee and a loyalty card — sensible answers that were already available before the meeting started.
The reason is associative: each idea is prompted by the previous one, so the group travels a short distance from where it began. Edward de Bono named the fix lateral thinking — entering the problem from a point you would not have reached by reasoning forwards from it.
A random object is the simplest possible entry point. Because it has no logical connection to the problem, your mind cannot reach it by normal association. Forcing a connection anyway produces genuinely new territory, and the object's irrelevance is precisely what makes it work.
The object is not the idea. It is a doorway into a part of the problem you would not otherwise have walked to.
The Basic Method
- State the problem in one sentence. Vague problems produce vague associations. "Increase weekday afternoon sales" beats "improve the business".
- Generate one object. Take the first result — rerolling for a "better" object defeats the entire mechanism.
- List the object's attributes. Not what it is, but what it does and what it is like. A ladder: extends, folds, is climbed, leans on something else, has discrete steps.
- Force a connection. Take each attribute and apply it to the problem, however absurd. Discrete steps → a tiered afternoon menu. Leans on something else → partner with the bookshop next door.
- Harvest, then judge. Capture everything first. Evaluation comes later, and never during generation.
Step three is the one people skip, and skipping it is why the technique sometimes fails. Jumping straight from "ladder" to "idea" leaves you searching for a metaphor. Listing attributes first gives you six or seven separate hooks instead of one.
A Forty-Minute Session Structure
| Time | Activity | Rule |
|---|---|---|
| 0–5 min | State and agree the problem | One sentence, written down |
| 5–10 min | Conventional ideas | Clear the obvious answers first |
| 10–15 min | Object one, attributes and links | No evaluation |
| 15–25 min | Objects two and three | No evaluation |
| 25–35 min | Cluster and combine | Look for hybrids |
| 35–40 min | Select three to develop | Now you may judge |
The second block matters more than it looks. Getting the obvious ideas out early stops them crowding the room later — people cannot think laterally while still holding an unsaid conventional suggestion.
Where the Technique Fits
- Product design — new features, form factors, unexpected use cases.
- Marketing — campaign angles and metaphors that competitors have not used.
- Writing — plot complications and character detail. See our writing prompts guide.
- Process improvement — reframing a workflow through an unrelated system.
- Naming — objects supply concrete imagery that abstract brainstorming rarely produces.
- Teaching — demonstrating creative method with a visible, repeatable procedure.
Common Mistakes
| Mistake | Why it fails | Fix |
|---|---|---|
| Rerolling for a relevant object | Relevance removes the lateral jump | Accept the first result |
| Judging while generating | Kills the absurd links that lead somewhere | Separate the two phases strictly |
| Skipping the attribute list | Leaves you hunting for one clever metaphor | Write six or seven attributes first |
| Using one object only | One doorway is easy to exhaust | Use three per session |
| Vague problem statement | Vague input, vague output | One specific written sentence |
A Worked Example
Abstract descriptions of lateral thinking are less convincing than seeing one run. Take a real problem: a local gym loses most new members within three months.
The conventional ideas arrive immediately — better onboarding, a discount for staying, an app. All reasonable, all already tried by competitors.
Now generate an object. Say it returns umbrella. The attributes: opens and closes, is carried but rarely used, protects from something, is easily forgotten, is often shared, is borrowed and not returned.
- Carried but rarely used → members pay for capacity they do not consume. Offer a low-cost tier that is explicitly for occasional use, removing the guilt that drives cancellation.
- Often shared → a membership two people can split across different days.
- Easily forgotten → leave something of the member's at the gym — a named locker, stored kit — so returning is the default.
- Protects from something → reposition the offer around what members avoid rather than what they gain.
None of these came from thinking harder about gyms. The guilt insight in particular is a genuinely different diagnosis of the retention problem, and it arrived through an umbrella.
Running It With a Group
Share a seed so everyone generates the same objects. This matters more than it sounds: if participants have different prompts, the session fragments into parallel solo exercises. With identical objects, the group can build on each other's associations, which is where the compounding value is. Seeds are explained here.
Have people write silently for two minutes before discussing each object. Silent generation before group discussion consistently produces more ideas and prevents the first confident voice from anchoring everyone else.
Frequently Asked Questions
Does this really work, or is it a party trick?
It is a recognised lateral thinking technique with decades of use in design and innovation practice. The mechanism — forcing association with an unrelated stimulus — is well established, though like any method it rewards being run properly.
How many objects per session?
Three is the practical sweet spot. One is too easy to exhaust; more than four produces volume without depth and leaves no time to develop anything.
What if no connection comes at all?
Return to the attribute list and work from a different attribute. If a connection genuinely will not form after two attempts, move on — not every object opens every problem.
Can I do this alone?
Yes, and it works well solo. Write rather than think — unwritten associations vanish before you can build on them.
Try It on a Real Problem
Write down a problem you are currently stuck on in one sentence. Open the Object Generator, take one object, and list six things it does. Then force each one onto your problem. The first three links will be useless; the fourth is usually where it starts working.
