<aside> 🛡️
+10% Revenue Per Visitor in 90 days. Or we refund the engagement. Book a teardown
</aside>
Your customers already wrote your best copy. It is sitting in survey exports, reviews, Reddit threads, and support tickets. This page is how you get Claude to mine it and hand back ad hooks, PDP headlines, and FAQ answers in the customer's own words.
Stop guessing what to say. Two recipes live here. One takes a survey export you already have and turns it into copy angles. The other scrapes your reviews and a competitor's reviews with Apify, then clusters them into desires, objections, and the exact phrases worth lifting for headlines. Same engine, two inputs.
The best hero headline you will ever run is one a customer wrote for you in a one-star review of a competitor. The language is already there. The job is extraction, not invention. Claude holds a couple hundred reviews in context at once, counts how often a theme repeats, and pulls the exact quote instead of paraphrasing it into mush.
Two inputs feed the same engine:
Both end in the same place: a ranked list of desired outcomes, objections, and verbatim phrases, each mapped to where it belongs (hero, FAQ, email flow).
The CSV is the data source, so the quality of the output tracks the quality of the export. Prep it right and the analysis is sharp.
| Brain | Hands | Outcome |
|---|---|---|
| Claude (Desktop or Code) | A survey CSV with descriptive headers | ICP summary, ranked buy-reasons and objections, verbatim quotes, plus 5 ad hooks and 3 LP headlines grounded in real language |
Any brand sitting on a post-purchase survey nobody reads. The chart in the survey dashboard is not the value. The value is the open-text answers, and nobody has time to read 800 of them by hand. This is the manual loop it kills: the founder or CRO who exports the survey, skims 30 responses, decides they have got the gist, and never touches the other 770.