Jake Thompson·
Fed 200 app-store reviews into this and got six ad angles ranked by awareness stage. Two of them beat our best performing creative in the first test round.
Extracts ad angles from real customer reviews and maps them to awareness stages with ready-to-test hooks.
Review-Mining Angle Matrix for Paid Social
You are a direct-response strategist who builds ad angles exclusively from customer language, never from marketing assumptions.
Product: {{product_description}}
Raw reviews (paste as many as you have):
{{reviews}}
Step 1 — Mine. Read every review. Extract verbatim phrases (quote them exactly, typos included) into four piles: pains before using the product, desired outcomes, surprising benefits the customer did not expect, and objections or near-cancellation moments.
Step 2 — Cluster. Group the phrases into 5-7 distinct angles. An angle is distinct only if it targets a different motivation — "saves time" and "less stress" are different angles even if the feature is the same. Name each angle in 3-5 words.
Step 3 — Matrix. Output a table with one row per angle: Angle name | Awareness stage (problem-aware / solution-aware / product-aware) | Strongest verbatim quote | Hook (rewrite the quote into a scroll-stopping first line, max 12 words) | Risk (why this angle might flop).
Step 4 — Rank. Order the angles by test priority using: frequency in reviews (how many customers said it) x emotional intensity of the language. Show the count next to each.
Rules: every hook must be traceable to a real quote — if you cannot point to one, drop the angle. Do not invent statistics. Flag any angle that would need a compliance check (health claims, income claims, before/after).
Ergebnisse
## Angle Matrix: FitStack (home workout app) — 212 reviews mined
### Verbatim piles (samples)
- Pains: "gym membership i used twice", "felt judged the second i walked in", "no idea if my form was right"
- Outcomes: "actually stuck with it for 3 months now", "20 min before the kids wake up"
- Surprises: "didnt expect the rest day reminders to matter but they do", "my knees stopped hurting"
- Objections: "almost cancelled when the trial ended", "thought it would be another app graveyard download"
### Matrix (ranked by test priority)
| # | Angle | Stage | Verbatim quote | Hook | Risk |
|---|-------|-------|----------------|------|------|
| 1 | Gym guilt refund (34 mentions) | Problem-aware | "gym membership i used twice" | "Your gym doesn't miss you. Your couch does." | Negative tone may fatigue fast |
| 2 | Sticking, not starting (29) | Solution-aware | "actually stuck with it for 3 months now" | "Month 3 is the real before-and-after." | Slower hook, needs strong visual |
| 3 | Judgment-free start (24) | Problem-aware | "felt judged the second i walked in" | "Nobody's watching. That's the point." | Overlaps angle 1 audiences |
| 4 | 20-minute parent window (19) | Solution-aware | "20 min before the kids wake up" | "Done before the first 'Muuum!'" | Narrows targeting to parents |
| 5 | Form confidence (16) | Product-aware | "no idea if my form was right" | "The app that tells you your squat is wrong." | Feature-led, weaker cold traffic |
| 6 | App graveyard skeptic (11) | Product-aware | "another app graveyard download" | "This one survives the app graveyard." | Meta-humor may miss broad audiences |
### Compliance flags
- "my knees stopped hurting" (8 mentions) — genuine cluster, but it is a health-outcome claim. Do not build an angle on it without legal review; excluded from the matrix.
### Test plan suggestion
Launch angles 1 and 2 head-to-head first: same audience, same format, hook is the only variable. Angle 4 goes straight to a parents lookalike — it will underperform broad but should win on segment.
Modell: GPT-4.1
17 Likes9 SavesScore: 12
1 Kommentar
Madison Carter·
The traceability rule is everything. If the hook can't point to a quote, it's fiction — and fiction fatigues.