Using public review data for product decisions
Reviews are free research
Every public review is a customer telling you, unprompted, what worked and what did not. Across retailers, marketplaces, app stores and maps, there are often far more reviews of a product than any survey would collect. The difficulty is that they are scattered and unstructured. Collecting them into one dataset turns them into research.
What to collect
- Rating and the site’s rating scale
- Review title and text
- Review date
- Verified purchase flag where shown
- Helpful votes count
- Product or location identifier, and the source site
- Language
Leave out reviewer names, profile links and photos by default. They are personal data and add nothing to product analysis. See public versus personal data.
Our reviews and ratings data service collects these from the sites you choose.
Use 1: Find what to fix
Group negative reviews by theme: battery, size, packaging, delivery, a specific feature. Count themes per product and per month. The top themes are a ranked list of problems in buyers’ own words. Teams often find one or two issues behind most complaints.
Use 2: Catch problems early
Track average rating and the share of low ratings each week. A sudden rise in one complaint after a supplier change, a formula change or an app release shows up in days. That is faster than waiting for returns or support tickets to build up.
Use 3: Learn from competitors
Collect reviews for competing products too. What do buyers praise? What do they complain about? Gaps in competitors’ products are positioning opportunities. Repeated praise tells you what the market values.
Use 4: Compare across channels
The same product can be rated differently on different sites, because buyers, packaging or delivery differ. Seeing ratings side by side shows whether a problem is the product or the channel.
Use 5: Locations and services
For restaurants, hotels, clinics and stores, ratings by location show where service differs. Operations teams can focus visits and training where ratings fall. See food and restaurants and travel and hospitality.
Apps
App store reviews spike after releases. Tracking ratings and review themes per version shows whether an update helped or hurt. See app store data.
Tips for analysis
- Use dates. Trends matter more than totals.
- Weight by recency. Old reviews may describe an old version.
- Read samples. Theme counts are a guide; read real reviews behind them.
- Watch for bursts. Sudden waves of very short reviews can be unusual activity, not genuine feedback.
- Keep it in context. Ratings are self-selected. Use them alongside your own data.
Getting started
Pick your top products and two or three competitors, and collect the full public history once. Then add new reviews weekly. For a sample, send us the products and sites or email [email protected].