Amazon review analysis: how to find what customers really complain about
Complaints are the most useful thing a competitor's customers can give you. They tell you what to fix, what to promise and what to show. But the way most people read Amazon reviews gives a distorted picture.
The problem with reading featured reviews
When you're not signed in, Amazon shows only a handful of featured reviews on a product page, typically 8 to 13, and full review pages require signing in. Featured reviews skew positive. In a test of 20 popular products, the featured reviews averaged fewer than one 1–3★ review per product. Scraping or reading them gives you a sample that hides most of the complaints.
Use "Customers say" instead
Many product pages now include a "Customers say" block: an AI summary that Amazon builds from all of the product's reviews. Alongside the summary, it lists topics, such as "Durability", "Odor" or "Battery life". Each topic is marked positive, negative or mixed, with how many customers mention it and how many of those mentions are positive or negative.
That turns review reading into numbers you can rank. Take one popular yoga mat:
- Durability: 941 mentions, 70% negative ("material cracked after a couple of uses")
- Odor: 464 mentions, 76% negative
- Stability: 610 mentions, mixed (some say it stays put, others that it slides)
- Quality, thickness and comfort: mostly positive, the reasons people buy it
The negative topics are a competitor's to-do list. The positive ones are table stakes you must match.
Read the whole category, not one product
One product's complaints tell you about that product. Complaints repeated across the top 20 tell you about the category, and that's where a new product can win. In the yoga mat category:
- Grip/slipping was a complaint on 11 of the 13 mats where it came up, including a $137 premium mat.
- Durability was the biggest complaint by volume, at 54% negative across about 8,800 mentions.
- Odor appeared on only 5 mats, but it was negative every time.
See the full yoga mat market report for the details.
How to turn complaints into a better listing
- Fix it, if you make the product. A charge indicator, a stronger strap or an odor-free material.
- Say it. Put the fix in the title or first bullet, in the buyers' own words ("no smell out of the box").
- Show it. Make one image per top complaint: a grip test, a tear test, the charger in the box.
- Check your own. Run the same analysis on your listing to catch problems before they show up in your rating.
Ask your AI: "What do customers complain about on [ASIN], and what should a competitor fix to win their buyers?"
Questions
Why not just scrape all the reviews?
Full review pages require signing in to Amazon, and scraping signed-in pages breaks Amazon's terms. The 'Customers say' summary is visible on the public product page and is built from all reviews, so it's both more complete and more appropriate to use.
Does every product have 'Customers say'?
Most established products do. New listings with few reviews often don't; in that case ZonRival falls back to the star-rating breakdown and the featured 1–3★ reviews, and says so.
Is 'Customers say' accurate?
It's Amazon's AI summary, so treat the counts as directional rather than exact. It's still far more representative than the 8–13 featured reviews.
Try it on your own competitors
ZonRival plugs into Claude and ChatGPT as a connector. Sign in with Google and you get 25 free credits, enough for a few full teardowns. Read-only; 12 Amazon marketplaces.