Twenty hooks and headlines, grouped by the hypothesis each one tests, written to the limits of the channel they will run on, and staged as ad variants when you are ready.
A headline test only teaches you something if the variants differ on purpose. This template produces a pack where every hook belongs to a labeled angle: curiosity, outcome, specific number, question, objection, contrast, identity. Within each angle Sprites writes a few executions, so a winning angle can be recognized even when one wording underperforms. The pack draws on your landing page, your reviews and the language your audience uses on Reddit and in competitor ads, so hooks sound like customers rather than like ads.
Every line is written to the destination: 40-character Meta headlines, 30-character Google responsive search ad headlines, first-three-seconds spoken hooks for TikTok and Reels, subject-line-length hooks for LinkedIn. The pack ships with a test plan, and the Meta or Google agent can stage the variants as ads in your account, each waiting for approval before it spends.
Sprites reads the landing page and pulls audience phrasing from reviews, Reddit and competitor ads in the Sprites Ads Library, so hooks use words buyers already say.
Six or seven angles are fixed up front. Each is one hypothesis about why someone stops scrolling, and the test is designed to compare angles before wordings.
Two to four executions per angle, each written to the destination format and limit, with a spoken version for video hooks and a shorter cut for Google headlines.
The pack is ordered into rounds: angles first, then the best executions within the winning angle, with a note on the sample needed to call a result.
Choose the ad set or ad group and Sprites stages one ad per hook with matching body copy. You approve the card before anything runs.
Run a round of six angle-level hooks against the control, read the winner, then run its executions. The optimization Autopilot proposes what to scale.
Get spoken opening lines for the first three seconds, the part of a video ad that decides most of its cost.
Feed a Google RSA with headlines that cover distinct angles, so ad strength improves for the right reason.
Reuse the winning ad hook as the page headline so message match holds from click to conversion.
Around twenty by default, grouped into six or seven angles with two to four executions each. Ask for more or fewer. The grouping matters more than the count: it lets a test show which reason to stop scrolling wins, not just which sentence.
Yes. Meta headlines, Google responsive search ad headlines and LinkedIn lengths are enforced per line, and video hooks are written to be spoken in about three seconds. Nothing in the pack needs trimming before it runs.
It stages the test. The Meta or Google agent creates one ad per hook in the ad set or ad group you choose and holds them in a review card for approval. Once live, the weekly optimization Autopilot reads the results and proposes which variants to pause or scale.
From your landing page and reviews, Reddit threads about the problem, and competitor ads in the Sprites Ads Library. Sprites quotes the source phrasing next to the hooks it inspired so you can judge the fit.
Yes. Sprites ships an official MCP server, so the same pack, test plan and staging flow work from Claude, ChatGPT, Grok or any MCP-compatible assistant.
Other prompts that run on the same agent. Browse the full gallery at /templates.
