Verified against Claude · 2026-07-31
Force a brainstorm into real, differentiated options instead of one idea repeated five ways
A constrained brainstorming prompt with hard kill criteria, mandatory diversity axes, and a required, sized tradeoff per idea, so options that violate a real constraint get cut outright and the ones that remain are genuinely comparable instead of cosmetic variations on the same idea.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Generate exactly 5 distinct ideas for the brief below. "Distinct" is defined precisely: each idea must differ from every other one on at least one of the axes under DIVERSITY REQUIREMENT. A rewording of the same core mechanism does not count as a second idea and will be rejected — check this yourself before presenting the list, not after I point it out. BRIEF Ways to get more of our existing email subscribers to actually open a Tuesday product-update newsletter, currently at a 12% open rate. SUCCESS METRIC Open rate on the Tuesday send, currently 12%, measured over the next 4 sends. Every idea should be evaluated, implicitly, against whether it plausibly moves this metric — an idea that's clever but disconnected from what actually matters here should be cut before it reaches the list, not included for variety's sake. DIVERSITY REQUIREMENT Across the 5 ideas, cover at least these different approaches, one idea per approach where the brief allows it: - Change to subject line or framing - Change to send time or day - Change to content format, not just wording - Change to audience segmentation - A structural change outside the email itself (different channel or cadence) HARD CONSTRAINTS — KILL CRITERIA No budget for new tooling. Must be implementable by one person in under a week. Cannot require re-platforming the email tool. Cannot reduce send frequency below once a week (contractual with sponsors). Any idea that violates one of these does not make the list at all, no matter how good it otherwise is — do not include it with a caveat that it "would need approval" or similar; if it fails a hard constraint, it is out. FOR EACH SURVIVING IDEA, STATE 1. The idea itself, in one or two sentences — concrete enough that someone could start building it tomorrow, not a direction or theme. 2. Which diversity axis it represents. 3. The one real tradeoff it makes, named specifically — not "requires more effort," but the actual thing given up (reach, speed, personalization, control) to gain the thing this idea is betting on. An idea presented with no real cost has not been thought through. IF AN IDEA SPANS MORE THAN ONE AXIS Pick the axis it represents most strongly and assign it there — do not let one idea count toward satisfying two axes at once just to shorten the list; the diversity requirement exists to force real spread across approaches, not to be satisfied on a technicality. 4. A rough size of the tradeoff, in the same units as the success metric where possible — if the metric is open rate, say roughly how much of a hit to some other number (send frequency, list size reached) this idea trades for it. AFTER THE LIST Name which single idea you would personally bet on and why, in two or three sentences — an actual opinion with a reason, not "it depends on your priorities." Then name the one idea on the list you'd cut first if forced to ship only three, and why — a ranked list, even an informal one, is more useful than a flat set with no ordering signal at all.
Customize
Optional — swap in your own details for the highlighted parts above.
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 — a well-documented pattern in unconstrained 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 repeatedly asking for 'another one.' Kill criteria that remove an idea outright, rather than including it with a caveat, matter because a model asked to generate creative options under loose guidance will often include an appealing idea with a soft hedge like 'would need approval' rather than actually enforcing the stated limit, which quietly reintroduces exactly the option the constraint was meant to rule out; making violation an exclusion rule rather than a caveat changes what actually appears in the output, not just how it's phrased. Sizing the tradeoff in the same units as the success metric turns a vague caveat into something comparable against the upside: a stated tradeoff with no rough size — 'costs some reach' — gives a decision-maker nothing to weigh, while a sized one lets two ideas actually be ranked against each other on the same axis the whole exercise cares about. The forced ranking at the end, both a bet and a cut, closes the same gap the tradeoff sizing addresses from the other direction: a neutral flat list defers the hardest part to the reader, and asking for both an endorsement and an elimination surfaces the model's actual implicit weighting from two angles at once, which also makes it harder to hedge into a wishy-washy 'it depends' as the final word.
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 — each with a named, sized tradeoff (e.g. "single-topic format: drops coverage of 2-3 secondary updates per week to gain a much stronger single hook, estimated worth 3-5 points of open rate based on similar switches elsewhere") — and a closing bet on the single-topic format, with the VIP-segment idea named as the first cut if only three ship.
Verified against
Claude Sonnet 4.6 · 2026-07-31
Changelog
- 2026-07-31 — 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
