Jake Thompson·
Dumped 60 customer reviews in and got seven ad angles ranked by how often the pain shows up — two are already in testing
Extracts ad angles from raw customer reviews, ranks them by frequency and emotional intensity, and drafts one hook per angle in the customer's own words.
Ad Angle Miner From Customer Reviews
You are a performance marketer mining customer reviews for ad angles. The best hooks are already written by customers — your job is to find and rank them, not invent new ones.
Product: {{product}}
Platform for the ads: {{platform}}
Reviews (raw paste, any format):
{{reviews}}
Process:
1. **Extract pain and outcome statements.** Go through every review and pull verbatim phrases describing (a) the pain before the product, (b) the moment of realization, (c) the outcome after. Keep the customer's exact words — typos and all.
2. **Cluster into angles.** Group the phrases into distinct angles (target 5–8). An angle = one pain + one promised outcome. Name each angle in 5 words or less.
3. **Rank.** Score each angle: frequency (how many reviews mention it) × intensity (how emotional the language is, 1–3). Show the math.
4. **Draft one hook per angle** for {{platform}}, using or closely echoing verbatim customer language. Mark quoted fragments with quotation marks.
5. **Flag the sleeper.** One low-frequency but high-intensity angle worth a cheap test.
Output as a table: rank | angle name | frequency | intensity | example verbatim | hook draft.
Rules: no angle may be based on a feature no reviewer mentioned. If reviews are too thin for 5 angles, say so instead of padding.
Ergebnisse
## Angle ranking — 58 usable reviews, 4 discarded (spam/blank)
| Rank | Angle | Freq | Intensity | Example verbatim | Hook draft |
|------|-------|------|-----------|------------------|------------|
| 1 | Sunday-night dread gone | 19 | 3 (57) | "i used to spend my whole sunday evening dreading the week" | "I used to spend my whole Sunday dreading Monday." 19 customers said some version of that — here's what changed. |
| 2 | Cancelled three other apps | 14 | 2 (28) | "finally canceled todoist, notion AND my paper planner lol" | She cancelled Todoist, Notion, AND her paper planner. One app survived. |
| 3 | Partner noticed the difference | 9 | 3 (27) | "my husband asked what got into me this month" | "My husband asked what got into me this month." It wasn't coffee. |
| 4 | Works with ADHD brain | 8 | 3 (24) | "first planner that doesn't fight my ADHD brain" | Every planner assumed a brain I don't have. This one didn't. |
| 5 | Setup in one evening | 11 | 1 (11) | "set the whole thing up during one episode of tv" | Set up your entire system in one episode of whatever you're watching. |
| 6 | Actually sticks past week 2 | 6 | 2 (12) | "month 4 and I still open it every morning, a record for me" | Month 4, still opening it daily. Her previous record: 11 days. |
| 7 | Cheaper than the guilt | 4 | 2 (8) | "costs less than the gym I never went to" | Costs less than the gym membership you're not using. Does more. |
## Sleeper pick
**Angle 4 (ADHD brain)** — only 8 mentions but every single one is intensity 3, unprompted, and uses near-identical phrasing ("doesn't fight my brain"). That phrase consistency across strangers is rare and usually means an underserved segment. Test with a $50/day cap on a narrow interest stack before scaling.
## Notes
- Angles 1 and 3 overlap emotionally (life-change proof). Don't run both in the same ad set — they'll cannibalize the same audience.
- No reviewer mentioned the AI features or templates gallery. Per your rules, no angle uses them — worth telling the product team that nobody's talking about the roadmap headliners.
- Review #41 ("changed my life" with no specifics) was excluded from scoring: high intensity, zero information.
Modell: GPT-4.1
44 Likes26 SavesScore: 31
2 Kommentare
Madison Carter·
Frequency times intensity with the math shown is exactly how this should be done. Too many angle docs are vibes.
Marcus Bell·
The note that nobody mentions the roadmap features is the real finding here. Product teams need to see that.
