| name | conceptual-exploration |
| description | Framework for exploring design breadth (structurally different approaches) and depth (pushing a chosen direction as far as possible). Use early in any design process when exploring options, brainstorming alternatives, or pushing a direction deeper. Trigger on: brainstorm, explore, alternatives, options, ideas, concepts, variations, what else, different approaches, push further, iterate, next level, how far can we take this, range, breadth, depth.
|
Conceptual Exploration
Two complementary practices: range (how broadly you explore before committing) and depth (how far you push a chosen direction). Range without depth produces scattered work. Depth without range produces polished versions of mediocre ideas.
Part 1: Conceptual Range
The danger of early commitment
Your first instinct is usually based on an assumption. Jumping to optimize it means you're refining one idea based on one assumption — and you might be missing a fundamentally better approach.
What breadth actually looks like
Variants of the same idea are NOT range. Three different layouts for the same concept are just skins. True range means structurally different approaches — each answering a fundamentally different question about how the problem could be solved.
Techniques to push past the obvious
The first 1-2 ideas come easily. Use these to expand:
- Remove or add a constraint. What if it didn't need a screen? What if it happened automatically? What if it worked offline?
- Blend from other domains. What if this was a game? A physical product? What would Muji make? What would a restaurant design?
- Invert the problem. Instead of helping users find what they want, help them eliminate what they don't.
- Set an arbitrary count. Force 5, 12, or 20 ideas. Later ones force unfamiliar territory where interesting ideas live.
- Optimize for a single facet. What if this was 10/10 on speed? On delight? On simplicity?
In practice
When asked to design or build something:
- Don't immediately build the first idea. Consider 2-3 structurally different approaches.
- Present the range with clear tradeoffs.
- Let the user choose or combine.
- Then go deep on the chosen direction.
If explicitly told to "just build it," still spend a moment considering whether the obvious approach is actually best. If you see a fundamentally better alternative, mention it briefly.
Part 2: Conceptual Depth
The spectrum (1 to 10)
- Level 1: The default. Minimum. Technically works.
- Level 3: Where most work ships. Obvious problems fixed, nobody pushed further.
- Level 5: Solid, professional. Meets industry standard.
- Level 7: Noticeably good. People feel quality even if they can't articulate why.
- Level 10: Best possible version. Everything considered, tried, edited, improved.
Most work stops between 1 and 3. Not because that's all possible, but because it's easy to stop pushing.
Different stages, different mindsets
- 1→3: Fix the obvious. Address gaps, edge cases. Resolution work.
- 3→6: Optimize and refine. Tighten spacing, improve hierarchy, simplify.
- 6→10: Discovery and invention. No longer clear what "better" looks like. Breakthroughs are unexpected.
How to push further
- Zoom in. Pick one element, give it all your attention. What if this one thing was world class?
- Remove something. Take away what isn't earning its place.
- Name what isn't working. "The spacing feels uneven" is more useful than "something's off."
- Reference the best. Find world-class examples. Notice what they do. Let them reveal gaps.
- Generate more. Sometimes the easiest way to improve is to create more variations and select the best.
The 70-iteration mindset
Great refinement takes more iterations than feels reasonable. Don't stop at iteration 3 and call it done. Ask: what would the next level look like?
The Complete Flow
- Explore range — Generate structurally different concepts
- Evaluate — Consider tradeoffs, identify strongest direction
- Commit — Choose with conviction
- Go deep — Push through progressive quality levels
- Know when to stop — Either you've hit 10, or found the limit and should reevaluate
