Protocol
Generate once before opening the visual examples
The first example you see can quietly become the shape of the answer.
When it fits
- You need references or competitor examples, but the task still requires genuinely different concepts rather than faithful imitation.
When to avoid it
- Examples can also provide useful knowledge, constraints and standards. The evidence does not support avoiding references entirely; it supports controlling when and how they enter the generation process.
Why it matters
Do one short independent generation pass before browsing screenshots, competitor pages or reference designs. Research on creative fixation shows that visual examples can reduce originality and can pull later solutions toward their features. The first pass does not need to be good; its job is to establish your own problem representation and several starting directions before the reference set narrows the search.
Steps
- Write the problem and hard constraints without opening the examples.
- Generate several meaningfully different mechanisms or layouts from memory.
- Only then inspect references for missing constraints, useful mechanisms and quality standards.
- Compare new ideas with the pre-reference list and note where the references pulled you toward copying.
An example
Before opening five competing landing pages, sketch three different ways to communicate the product's value. Then use competitors as evidence about conventions and gaps rather than as templates.
Check your result
At least one candidate direction existed before example exposure, and you can tell which later features came from the references.
Keep this limit in mind
- Examples can also provide useful knowledge, constraints and standards. The evidence does not support avoiding references entirely; it supports controlling when and how they enter the generation process.
Connected ideas
Use beforeSeparate generation, clarification and ranking
Evidence and sources
A 2020 study found that visual example exposure reduced originality on a creative idea-generation task; explicit instructions to avoid the example helped mitigate this fixation, while verbal examples with avoid instructions increased originality in one experiment.
Effects depended on example modality and instruction; the study does not justify removing all examples from creative work.
Earlier laboratory design experiments found that nonexperts tended to reproduce inappropriate elements from pictorial examples, and explicit defixating instructions reduced the effect.
The tasks were controlled design problems with nonexperts and do not estimate fixation magnitude for professional design systems.