Verified against Claude · 2026-07-16
Force a brainstorm to produce real options, not ten flavors of one idea
A brainstorming prompt that imposes hard structural constraints — a fixed count, mandatory diversity across a named axis, and a stated tradeoff per idea — so Claude produces genuinely distinct options instead of minor variations on its first idea.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Generate exactly 5 distinct ideas for the brief below. "Distinct" means each idea must differ from every other one on at least one of the axes listed under DIVERSITY REQUIREMENT — small variations on the same core idea do not count as separate ideas and will be rejected. BRIEF Ways to get more of our existing email subscribers to actually open a Tuesday product-update newsletter, which currently gets a 12% open rate. DIVERSITY REQUIREMENT Across the 5 ideas, cover at least these different axes, one idea per approach where possible: - Change to subject line / framing - Change to send time or frequency - Change to content format (not just wording) - Change to audience segmentation - A structural change unrelated to the email itself (e.g. a different channel) HARD CONSTRAINTS No budget for new tooling. Must be implementable by one person in under a week. Cannot require re-platforming the email tool. FOR EACH IDEA, STATE 1. The idea itself, in one or two sentences. 2. Which diversity axis it represents. 3. The one real tradeoff it makes — every idea gives something up to gain something else; name what this one gives up. An idea with no stated tradeoff has not been thought through yet. AFTER THE LIST Name which idea you would personally bet on and why, in two sentences — a real opinion, not "it depends on your priorities."
Customize the highlighted detailsoptional — the prompt above already works
Why this works
Left unconstrained, a 'give me ideas' prompt reliably produces a cluster of near-identical suggestions dressed in different words, because the model's first plausible idea sets a gravity well that later ideas drift toward rather than escape — this is a well-documented pattern in unconstrained LLM brainstorming, not specific to any one topic. Naming explicit diversity axes and requiring each idea to map to a different one forces the search outward across the solution space instead of deeper into the first branch found, the same mechanic that makes a lateral-thinking prompt more productive than a repeated 'give me another one.' Requiring a stated tradeoff per idea is the constraint doing the most real work here: a brainstorm with no cost attached to any option is not actually useful for a decision, since every real option trades something off, and forcing the model to name it prevents the common failure of ideas presented as strictly better with no visible downside, which is rarely true and hides the actual decision a reader needs to make. Ending with a forced personal bet — not 'it depends' — matters because a brainstorm that stops at a neutral list leaves exactly the hardest part, choosing, entirely to the reader; asking Claude to commit to an opinion surfaces its actual implicit weighting of the tradeoffs instead of hiding it behind false even-handedness.
What you get back
5 ideas spanning subject-line reframing, a switch to Thursday send, a shift from digest-format to single-topic emails, a VIP-segment early-access version, and a move of the update to a short in-app notification instead of email — each with a named tradeoff (e.g. "single-topic format: gives up covering multiple updates per week, gains a much stronger single hook"), and a closing bet on the single-topic format as most likely to move the needle without new tooling.
Verified against
Claude Sonnet 4.6 · 2026-07-16
Changelog
- 2026-07-16 — Initial publish, verified against Claude Sonnet 4.6.
Building this for real?
This is a free starting point. If you'd rather have what Scult builds built and running for your business, that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
