Turn reviews into ad hypotheses by preserving each source, clustering repeated customer language, separating observation from interpretation, checking for counter-evidence, and writing one narrow message hypothesis the market can test.
1. Preserve the source before summarizing it
Capture the review URL or source reference, date, product or variant, exact language, and any context needed to understand the statement. A summary without a source is harder to verify and easier to overstate.
Do not treat a review as permission to make a regulated, medical, comparative, or guaranteed claim. The brand remains responsible for substantiation and compliance.
2. Normalize without erasing meaning
Customers describe the same situation with different words. Normalize spelling and obvious duplicates, but keep the original phrasing available. The language itself may reveal the customer's awareness, desired progress, failed alternative, or objection.
- Situation: When or why was the product considered?
- Desired progress: What did the customer want to become easier or better?
- Friction: What felt slow, confusing, expensive, risky, or disappointing?
- Alternative: What had the customer tried before?
- Proof language: What made the outcome believable to that reviewer?
3. Build patterns, not cherry-picked quotes
Group comments that describe a similar situation or belief. Record how many relevant sources support the pattern, which sources disagree, and whether the sample is biased toward a product, channel, or unusually vocal customer group.
A large cluster can justify attention. It does not automatically justify a universal statement. Write recurring pattern when that is what the evidence shows.
4. Separate observation from interpretation
This separation is what keeps the strategy honest. The review supports the observation. The strategist owns the interpretation. The live campaign resolves the hypothesis.
Observation: several reviews say the existing routine has too many steps. Interpretation: simplicity may be a meaningful decision criterion. Hypothesis: a three-step demonstration for routine-fatigued shoppers will earn more qualified attention than a broad lifestyle message. The hypothesis is a proposed test, not a proven fact.
5. Add product truth and proof boundaries
A customer pattern becomes a viable direction only if the product can address it and the brand can support the message. Connect the pattern to an accurate product mechanism, demonstration, specification, or other available proof.
If the product evidence is weak, narrow the claim, gather more support, or keep the direction exploratory. Do not borrow certainty from emotional customer language.
6. Write the hypothesis and intended learning
Name the customer, message, evidence, expected response, and the decision the team will make after the test. Keep the direction distinct from the other directions in the pack.
Common mistakes
- Using one vivid review as if it represents the market
- Rewriting customer language until the original meaning disappears
- Ignoring negative or contradictory reviews
- Turning an outcome anecdote into a guaranteed claim
- Generating quotes that no customer actually wrote
- Testing several messages at once and learning nothing specific
Frequently asked questions
How many reviews are needed before writing an ad hypothesis?
There is no universal minimum. Use the available sample honestly, record its limits, and lower certainty when the pattern is supported by only a few relevant sources.
Can customer reviews be quoted in ads?
That depends on permissions, platform rules, applicable law, and the nature of the claim. Preserve the source and obtain legal or compliance review when needed.
What is the difference between an insight and a hypothesis?
An insight is an interpretation of evidence. A hypothesis predicts how a specific message or direction may affect a test and can be challenged by the result.
Should contradictory reviews be removed?
No. Counter-evidence helps the team understand where a pattern may not apply and prevents an overly broad claim.