Verified against ChatGPT · 2026-08-10
Generate a set of testable product title variants instead of one guess at the best title
Produces 5 structurally distinct product title candidates, each optimized for a different plausible buyer trigger, so you have a real A/B test set instead of five reworded versions of the same idea.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Give me 5 product title variants for Hyprflow reusable water bottle, 32oz that I can actually A/B test against each other — not five versions of the same title with synonyms swapped, but five titles built on genuinely different hypotheses about what makes a buyer click. PRODUCT AND KEY ATTRIBUTES Vacuum insulated, keeps cold 24hrs, leak-proof straw lid, fits standard car cup holders, made from recycled stainless steel PLATFORM THIS TITLE WILL RUN ON Shopify collection page and Google Shopping feed CHARACTER LIMIT 70 characters for Google Shopping title field COMPETING TITLES ON THE SAME RESULTS PAGE "HydroPro Insulated Bottle 32oz" and "EcoSteel Water Bottle - BPA Free" Build each variant around a different trigger, and label which trigger it's testing: Variant A — SPECIFICITY: leads with the single most concrete, differentiating attribute (a number, a material, a mechanism). Variant B — USE CASE: leads with who it's for or when it's used, not what it is. Variant C — COMPARISON IMPLICIT: structured to read as a clear upgrade over what the competing titles on the page are offering, without naming a competitor. Variant D — OUTCOME: leads with the result the buyer gets, with the product category named second. Variant E — MINIMAL: strips to the fewest words that still fully identify the product, betting that clarity beats persuasion at the title stage. Each variant must respect the character limit exactly and must not repeat the same opening word as another variant — if two variants would naturally start the same way, force a genuine structural difference instead. WHAT NOT TO DO Do not use any variant as a dumping ground for every attribute — a title is not the place to list everything; each variant tests one hypothesis and should stay disciplined to it even if that means leaving out a true, positive fact. OUTPUT FORMAT A table: variant letter, trigger name, the title text, character count, and one sentence on what result would tell you this variant won (e.g., "higher CTR would mean buyers respond to concreteness over outcome framing").
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Asked for "five title options" without structural constraints, GPT-5.1 tends to produce five paraphrases clustered around the same underlying idea, because the model is sampling from the same region of its output distribution each time rather than being pointed at genuinely different strategies — naming five distinct psychological triggers up front forces the sampling into different regions and produces variants that are actually informative to test against each other, rather than five near-duplicates that would just split traffic without teaching you anything. Requiring each variant to name what winning would prove is what turns this from a copywriting exercise into an actual experiment design — without that line, a merchant running an A/B test on five similar-looking titles has no way to interpret a result beyond "this one word choice happened to work," which doesn't generalize to the next product. The explicit instruction against dumping every attribute into each variant addresses GPT-5.1's tendency, under a hard character limit, to try to maximize information density rather than commit to one clear angle — a title that tries to be concrete, use-case-driven, and outcome-focused simultaneously within 70 characters usually ends up saying nothing clearly, which defeats the point of testing hypotheses in isolation.
What you get back
A (Specificity): Hyprflow 32oz Steel Bottle, Cold 24 Hrs, Leak-Proof Straw — 54 chars — tests whether the cold-retention number alone drives clicks. B (Use Case): Hyprflow Bottle for Commutes — Fits Any Car Cup Holder — 58 chars — tests whether context beats spec...
Verified against
ChatGPT GPT-5.1 · 2026-08-10
Changelog
- 2026-08-10 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
