Reading Amazon data correctly: estimates, pooled figures and traps
Reading Amazon data correctly: estimates, pooled figures and traps
Which numbers are facts and which are estimates, what a parent listing pools across variations, what a rank is not, and the mistakes that make a reasonable-looking analysis wrong.
Amazon Review Scraper vs a Reviews API
Four ways to scrape Amazon reviews, from a browser extension to your own Python scraper, a scraping API and a reviews data API, and which one fits which job.
Amazon Backend Search Terms: What Goes in the Field and How to Fill It From Reverse Lookup Data
The backend search terms field is short, counted in bytes, and ignored entirely when it runs over. This post covers the rules Amazon publishes for it, why nobody can read a competitor's backend field, and a six-step way to fill yours from the terms competitors already rank for.
How to Check Amazon Keyword Search Volume: Four Sources and What Each Number Means
Amazon does not publish a search volume figure for every keyword. Four sources give you something close: two Brand Analytics reports, Product Opportunity Explorer and third-party estimates. This post shows what each one returns, who can use it and how to pull the number through an API.
Where an Amazon Listing's Visibility Comes From: Search, Ads and Related-Product Placements
A product is seen in two places: on search results pages and on other products' detail pages. Two API calls count both for any ASIN. In our test, one product appeared for 1,237 keywords organically and 190 through ads, and sat on 1,264 related-product placements, 21 of them paid.
Reading Negative Amazon Reviews by Variation: Which Complaints Belong to Which Product
A variation family pools its reviews, so the complaints on one page can belong to several different products. In a sample of 20 recent critical reviews we pulled, 3 were about a spare part sold as a variation, and the one-star and three-star reviews described two different problems. This post shows how to split them.
Parent and Child ASINs: Which Numbers Are Pooled Across Variations and Which Are Not
A parent ASIN is a shell that groups variations; a child ASIN is the thing a shopper buys. Reviews, sales estimates, price and BSR mean different things at the two levels. This post shows which is which, how to tell the levels apart in API data, and how to keep comparisons on one level.
Choosing a Price Point on Amazon: Read the Category's Price Bands Before You Set a Price
In one category we tested, 36 of the top 100 products sat between $15 and $20 and took 42% of unit sales, while the four products at $20 to $25 each sold more than any other band. This post shows how to read a price distribution into three numbers per band and what they say about where to price.
Searching Amazon Products by API: Match Types, Sort Order, and Why It Is Not the Search Results Page
One keyword returned 265,533 products on broad match, 1,880 on phrase and 750 on exact. This post shows what each match type does with a real test, how results are ordered, and when you need search positions instead of a product search.
Getting Amazon Reviews Through an API: What the Official APIs Return and What They Leave Out
None of Amazon's own APIs returns the review text of an arbitrary product. This post goes through what the seller, affiliate and advertising APIs do return about reviews, where the gap is, and what a request for review text by ASIN looks like.
Is There a Google Trends API? The Official Alpha, the Alternatives and What Each Returns
Google announced an official Trends API in July 2025. As of October 2026 it is still an alpha you apply for, with no self-serve key and no published price. This post covers what the official API returns, what the other routes are, and how Amazon sellers use Trends data as a cross-check.
Amazon Brand Analytics Search Terms: How to Read Rank, Click Share and Entry Cost
ABA gives a search frequency rank, not a volume. One row tells you three things: which way the rank is moving (a negative change means worse), how much of the clicks the top three ASINs hold, and what reaching page one costs.
Choosing an Amazon Data Analysis Tool: Test It With Data You Already Know
Every tool looks accurate in a demo. What separates them is comparing their estimates against ASINs whose real sales you hold in Seller Central — reading direction, magnitude and ranking — and then asking a few questions about definitions.
Which Amazon Data You Can Get for Free, and Which You Cannot
Free Amazon data is real, and most of it is accurate. Its boundary is clear: what is free is mostly 'now' and 'yours', and what costs money is almost always 'the past' and 'someone else's'. This post lists it by source and explains why this site offers only a short trial and no free tier.
Choosing an Amazon Keyword Tool: First Work Out Which "Search Volume" It Shows
"Which keyword tool is most accurate" usually compares two different quantities. A single reverse lookup response holds four fields that look like search volume, each with its own period and scope. This post splits the keyword toolkit by task and lists the three questions to ask when evaluating one.
Product Research on Amazon US, Japan and Europe: The Same Analysis Does Not Transfer As-Is
Product research methods carry across marketplaces; the numbers do not. Every request needs a marketplace (four common endpoints silently default to the US), amounts come in each marketplace's currency, keywords are stored in its language and category paths are defined per marketplace. This post sets out what to redo and what to reuse.
Building an Amazon data report: if it cannot be compared to last week, it is not a report
Reports go unread not because they lack detail but because they cannot be set beside the previous one. The three things that must stay fixed, a minimum report structure, and three habits that void a report.
Amazon category analysis: four distributions decide whether you can still get in
A big category total does not mean a share of it is available to you. Listing age, seller type and price distributions, plus one statistics call, read the competitive structure — above all whether new listings still have room.
Amazon data analysis: separate facts, estimates and your own judgment first
The expensive mistake in Amazon data analysis is not miscalculating — it is using an estimate as if it were a fact. The boundary between three kinds of data, which to query for which question, and one recording rule that applies to all of it.
Where Amazon data comes from: why there is no single "Amazon database"
People searching for a database want a dataset to download. Data splits along authorization boundaries into three blocks, and it is a stream rather than a store. A map of what exists, how fast each field goes stale, and whether you can keep a copy.
Amazon Market Analysis: Where the Market Size Number Comes From
Market size is the least reliable number in the dataset — it rests on a head sample and two layers of estimation. This post opens up how it is computed, the unit trap in ratio fields, and how to read share properly.
Choosing an Amazon product research tool: the difference is how much it decides for you
What separates product research tools is not whose data is more complete — it is who set the weights behind that "opportunity score". Three tool shapes, an acceptance checklist you can run, and how to write your own criteria instead.
How to analyse Amazon reviews: from a pile of them to three kinds of actionable conclusion
The output of review analysis should not be a summary. It should be three separated conclusions — fix the product, fix the listing, fix nothing — because blending them turns a description problem into an expensive product change.
Amazon Review Types Explained: Verified Purchase, Vine, Free Product and Early Experience
What each review label means, how to tell them apart in API data (verified, vine, free, experience), and why a star average computed before separating them has no commercial meaning.
What Amazon sales data can and cannot tell you: the API labels its own estimates
The marketplace never publishes unit sales, so every monthly figure is derived from BSR. This breaks down the fields across three sales endpoints — which are facts, which are derived, and the parent-versus-child comparison that quietly inflates results.
What Is an ASIN? vs UPC, EAN, ISBN, SKU
An ASIN is the ten-character code Amazon gives every product. How it differs from UPC, EAN, ISBN, SKU and FNSKU, how book ASINs work, and what it opens up.
Why General-Purpose LLMs Get Amazon Data Wrong, and Three Limits That Remain After You Connect Real Data
A model inventing ASINs, prices, and sales figures is structural, not a glitch. Here is why — and the three boundaries that still apply once a real data API is wired in.