Done-for-you research report
What buyers in your category complain about — and who promises the opposite
We read the reviews across 50–300 products in your category and check every theme against what those same listings actually claim. You get a report naming the gaps, and the products whose own copy contradicts their buyers.
Public marketplace data. No account, no integration, nothing to install.
Fabric feels scratchy
Sizing runs small
Machine washable
Illustrative — not a published result
Three kinds of finding, one analysis
Reading reviews tells you what customers say. Reading listings tells you what sellers claim. The value is in the overlap — and it only exists if you hold both against each other, product by product.
Illustrative figures, shown to explain the format — not results from a published study.
The product promises it. The reviews deny it.
Products that make a specific claim and carry reviews reporting the opposite. Matched per product, so the claim and the complaint belong to the same listing — never two separate category percentages placed side by side.
23% of sampled products both advertise “soft and skin-friendly” and have reviews reporting a scratchy fabric.
It comes up across the category. The listings stay quiet.
Themes that recur across many products in the sample while the listings barely address them. Often answerable with a content change — though not always: some blind spots point at the product, not the copy.
54% of products have reviews raising sizing. 11% of those listings mention fit at all.
The listings push it. The reviews rarely bring it up.
Features the category features prominently that seldom appear in review text. Worth testing rather than cutting — silence often means a feature is simply expected — but it shows where attention and interest may be mismatched.
79% of listings feature machine-washability. It appears in reviews for 3% of products.
What a finding looks like
| Theme | Products with reviews on it | What the same products’ listings do | Finding |
|---|---|---|---|
| Fabric feels scratchy | 41% | 23% of sampled products both claim “soft” and carry these reviews | Contradiction |
| Sizing runs small | 54% | 11% of those listings mention fit at all | Blind spot |
| Machine washable | 3% | 79% of sampled products feature it prominently | Possible overinvestment |
Denominators: the review column is the share of products in the sample whose reviews contain the theme at least once. The listing column is stated per row, and a contradiction is always counted on the same product — one listing that both makes the claim and carries the conflicting reviews. Every table in a report states its sample size, marketplace and snapshot date.
How a category analysis runs
01
You name the category
A marketplace category, a search term, or a list of competitors you want treated as the set.
02
We agree the sample
Typically 50–300 products, snapshot-dated — selected by sales rank on Amazon, or from an agreed collection or competitor list on Shopify, where no public ranking exists. We confirm the boundary with you before any analysis runs, so we are measuring the category you mean.
03
We analyse the reviews
Themes are extracted from the review text we can retrieve for every product in the sample, clustered, and counted with sentiment per theme — not one blended score per product.
04
We cross-check the listings
Each theme is matched against the same product’s own listing content — bullets and A+ content on Amazon, the product page copy on Shopify. That is where contradictions, blind spots and possible overinvestment surface. Where listing content is unavailable, we report it as unexamined rather than as missing coverage.
05
You get a preview
The top themes, the sample definition, and the strongest matched finding if the analysis produced one. An honest report is allowed to find no contradictions, and we will say so.
What you get back
Preview
Sent after we agree the sample
Top themes by frequency across the sample
The strongest matched finding if there is one, with the products named
Sample size, marketplace and snapshot date
The method, written out, so you can challenge it
Full report
Quoted once the sample is set
Every theme, ranked, with its sentiment split
Every contradiction and blind-spot finding the analysis produced, matched per product
Per-product coverage breakdown
The phrasing buyers use for each theme, in their own words
Data export
Optional re-run at an agreed interval to track movement
Who asks for this
Brands entering a category
What buyers already complain about, before you commit to a product spec or a positioning line.
Sellers with a catalogue
Which of your own listings make a claim your reviews contradict — the kind of gap between promise and experience that can contribute to returns and poor ratings.
Agencies
Category evidence for a pitch or an audit, with the products named and the method documented.
Product and research teams
What buyers raise most across a category, as input to a decision you still own.
What this does not do
Worth knowing before you ask, so the report answers a question it can actually answer.
It is review data, not a survey programme
We analyse public marketplace reviews. No surveys, support tickets, interviews or NPS. If you need feedback from your own customers across channels, you need a voice-of-customer platform, not this.
Reviewers are self-selected
People who chose to write a review are not a representative sample of everyone who bought. This is a strong signal of what buyers discuss publicly — which is what future buyers read — not a measurement of all customers.
Products are not equally exposed
A product with thousands of retrieved reviews has more chances to register a theme than one with ten, so “share of products mentioning it” is not a like-for-like comparison. Every report states how many reviews were retrieved per product, the time window, and anything we could not access.
Frequency is not importance
A theme appearing across half the category matters, but the report ranks what is said, not what drives revenue. Which findings are worth acting on stays your call, and we will say so in the report rather than imply otherwise.
A match is evidence, not a verdict
When a listing claims “drip-free” and its reviews report dripping, we report a mismatch to investigate. It may be a false claim, a changed product, or a misread. We name the products so you can check.
Questions people ask
What is customer feedback analysis?
Systematically reading customer feedback to find recurring themes, instead of reading reviews one at a time. We run it across an agreed sample of a whole product category rather than a single product, so the result is a category benchmark rather than one product’s to-do list.
How is this different from voice of customer software?
Voice-of-customer platforms help you collect and track feedback from your own customers over time, typically including surveys and metrics like NPS. We analyse reviews that already exist publicly, across a whole category including competitors, and cross-check them against what those listings claim. Different input, different question — and for feedback from your own customers across channels, a VoC platform is the right tool, not this.
Do I have to give you any data?
No. We work from public marketplace data. There is no integration, no account and nothing to install.
Which marketplaces do you cover?
Amazon product reviews, and Shopify store reviews through Judge.me.
How big is a category?
Typically 50–300 products. We agree the sample with you before the analysis runs and state it in the report, because the category boundary changes the result.
What does a percentage in the report mean?
Unless stated otherwise, it is the share of products in the sample whose reviews contain that theme at least once. Contradictions are matched per product: the same listing makes the claim and carries the conflicting reviews. Every table states its denominator, sample size and snapshot date.
Can you analyse my competitors?
Yes, and that is usually the point. Any product we can retrieve reviews for on the marketplaces we cover — Amazon, and Shopify stores using Judge.me — can be included in the sample.
What if you find nothing interesting?
Then the report says so. A category where listings broadly match what buyers say is a real finding, and we would rather tell you that than manufacture a contradiction to justify the work.
What does it cost?
We quote once the sample is agreed, because cost follows category size and review volume rather than a fixed package.
Which category should we look at?
Tell us the category or send one example product. We confirm the sample with you, run the analysis, and send the preview.
We reply within two working days with a scope confirmation — no analysis starts before you have agreed what gets measured. The preview is free once the sample is agreed; the full report is quoted.
Prefer email? Write to support@descripio.com