Verified against ChatGPT · 2026-07-21
Generate title options that raise CTR without promising something the video doesn't deliver
Produces a batch of title candidates scored against a stated click-through hypothesis and cross-checked line by line against what the video actually contains, so no option survives that the thumbnail-title pair would have to mislead on to earn the click.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are generating YouTube title options for a finished or near-finished video. Every title must be something a viewer would feel was accurate after watching the whole video — you are optimizing for click-through rate that survives contact with the actual content, not first-impression curiosity that collapses into a dislike or a fast exit once the video doesn't deliver what the title implied. VIDEO CONTENT SUMMARY A personal-finance video testing four popular budgeting apps against the same real bank data for one month; two apps missed recurring subscriptions entirely, one had a data sync bug that double-counted a paycheck CURRENT OR PLACEHOLDER TITLE Testing 4 Budgeting Apps TARGET KEYWORD OR SEARCH INTENT best budgeting app 2026 THUMBNAIL CONCEPT (IF DECIDED) Thumbnail shows four app logos with a big red X over two of them CHANNEL VOICE Dry, slightly skeptical, numbers-first — avoids exclamation points and hype language TITLE RULES Every specific number, superlative, or claim in a title — "5 mistakes," "the worst," "nobody tells you" — must be verifiably true of the video content summary provided; do not include something the video does not actually contain to make a title stronger. If the target keyword or search phrase is provided, at least half the title options must contain it in a natural position, since search-driven traffic depends on that literal match appearing in the title, not just implied by synonyms. Where a thumbnail concept is already decided, the title and thumbnail must not both make the same point — if the thumbnail already shows the shocking result visually, the title should supply the missing context or stakes instead of restating what the image already says, since a title-thumbnail pair that duplicates the same information wastes one of the two slots a viewer actually reads before deciding to click. Vary the mechanism across the options rather than producing five versions of the same curiosity-gap phrasing: include at least one option built on a stated number, one built on a direct question, one built on a specific named mistake or contradiction, and one that states the outcome plainly with no gap at all, since a fully transparent title is sometimes the highest-CTR option for a viewer who already trusts the channel and doesn't need to be teased. Match the channel voice given — a title that reads as a different channel wrote it breaks the trust a returning subscriber has already built with this specific channel's tone. OUTPUT FORMAT A numbered list of at least eight title options, each under 60 characters where possible. After each title, one line naming its specific mechanism (number, question, contradiction, plain statement, etc.) and one line confirming which exact part of the video content it is drawn from. Flag any title you considered but rejected specifically because it wasn't fully supportable by the content, and say why.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Click-through rate on impressions is the metric YouTube Studio surfaces as the direct measure of how well a title and thumbnail pair convert an impression into a view, but CTR earned by a claim the video doesn't back up shows up almost immediately as a collapsed audience-retention curve and a spike in the explicit dislike rate, because the viewer who clicked feels misled within the first minute — which is why this prompt fact-checks every number and superlative against the actual content summary before a title is allowed to survive, rather than treating CTR as the only variable that matters. Requiring the target keyword to appear literally, not just implied, in at least half the options reflects how YouTube's search indexing actually weighs title text as one of its strongest ranking signals for a search query, distinct from and generally stronger than the description or tags — a title that captures the idea of "best budgeting app 2026" through a clever paraphrase will not surface for that literal search the way a title containing the phrase itself will. The rule against the title and thumbnail restating the same information targets a specific, measurable inefficiency: a viewer scanning a results page or a suggested-videos rail processes the thumbnail and title together in under a second, and if both slots convey the identical fact — the image already shows the shocking result, the title just says it again in words — one of the two highest-value pieces of real estate a creator gets is being wasted on redundancy instead of adding context, stakes, or a second distinct hook. Deliberately varying the mechanism across options, including requiring one fully transparent, no-curiosity-gap option, matters because CTR performance is genuinely audience-dependent: a cold, first-time viewer from search often responds best to a plain, specific statement of outcome, while a returning subscriber already primed to trust the channel will click a vaguer curiosity-driven title precisely because they don't need to be convinced the payoff is real — testing across mechanisms rather than one formula is what actually surfaces which lever this specific video and audience responds to. Matching channel voice matters for a returning-viewer signal the algorithm also tracks: a title that reads like a different, more sensational channel wrote it can win a single click while quietly training regular viewers that this channel's titles can no longer be trusted at face value, which erodes exactly the return-viewer trust that drives long-term channel growth more reliably than any single video's CTR.
Verified against
ChatGPT GPT-5.1 · 2026-07-21
Claude Sonnet 4.6 · 2026-08-01
Changelog
- 2026-07-21 — Initial publish, verified against ChatGPT (GPT-5.1) and Claude (Sonnet 4.6).
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
