The best-known list of top Amazon marketplace sellers is published by Marketplace Pulse, and it does not rank sellers by sales. It ranks them by how many seller reviews each storefront collected in the last 30 days. The page says so in its own words: “Seller rank is based on the number of reviews each seller received during the last 30 days, which counts feedback received from shoppers after a purchase” (data checked 2026-09-06).
That single sentence is the whole story, and it is the part every article that reposts the list leaves out. A review count is a proxy for sales volume, not a measurement of it. This guide takes the proxy apart — where it comes from, the four conditions under which it distorts, two things the live page does that it does not advertise, and a checklist for reading any “biggest Amazon seller” list you run into.
What “top Amazon marketplace sellers” lists actually count
The metric is seller feedback, not product reviews. Those are two different objects on Amazon. A product review attaches to an ASIN and can be left by anyone who bought that item from any seller. Seller feedback attaches to the storefront and rates the transaction — the buyer’s path, the three-day wait before the button appears, and the 90-day window are covered in our guide to submitting Amazon seller feedback.
Marketplace Pulse describes itself as “an e-commerce intelligence firm” whose “proprietary software parses raw data from web-scraping and APIs to extract critical pieces of structured data on brands, products, categories, and sellers” (checked 2026-09-06). Nothing in that description involves Amazon handing over sales figures, because Amazon does not hand over per-seller sales figures to anyone.
So the chain runs: shopper buys → some fraction of shoppers leave feedback → that fraction is counted → the count is used to order the list. Marketplace Pulse states the leakage in that chain explicitly on the same page: “Providing feedback is optional, and the percentage of customers who leave a seller review is anywhere in the range of 1–5%, depending on the product’s pricing and category” (checked 2026-09-06).
Read that carefully. A 1% response rate and a 5% response rate are a fivefold difference in how many orders one review represents. The list itself tells you the conversion factor is unknown and category-dependent. It is a ranking of feedback volume that correlates with sales volume, and the source says as much.
The public list on 2026-09-06
Here is what the free Amazon.com page showed. The page carried the note “updated on September 5th” and listed 100 rows, with tabs for the USA, UK, Germany, France, Canada, Japan, India, Italy, Spain, and Australia.
| Rank | Seller | Country | 30 Day Reviews | Lifetime Reviews |
|---|---|---|---|---|
| 1 | Pattern. | United States | 5,438 (-5%) | 517,349 |
| 2 | eSupplements | United States | 3,314 (-8%) | 237,331 |
| 3 | ZEAGOO | United States | 2,835 (-2%) | 34,851 |
| 4 | Carlyle | United States | 2,636 (-12%) | 266,229 |
| 5 | Bayland Health | United States | 2,564 (1%) | 80,663 |
Source: Marketplace Pulse, Top Amazon Marketplace Sellers, list dated September 5th, retrieved 2026-09-06. The basis is 30-day seller feedback counts, not sales.
One column is missing from the table above on purpose. The page also prints a Yearly Revenue column, and every value in it sits behind a lock icon — it is a paid field. More on that below, because what it does while locked is worth knowing.
Where “reviews as a sales proxy” breaks
The proxy is not wrong. It is narrow. Four conditions push a storefront up or down the list for reasons unrelated to how much it sold.
Price is invisible. The count is per order that generated feedback, and an order is an order. A storefront moving $9 phone cables and a storefront moving $400 tool sets can post identical 30-day review counts while their revenue differs by a factor of forty. The list has no price weighting and does not claim to. This is why a feedback ranking and a revenue ranking are genuinely different objects, not two views of one thing — see how much Amazon sellers make for the revenue side and the sources that actually model it.
Category feedback rates differ. This is the 1–5% band Marketplace Pulse names. Cheap consumables and books convert differently from electronics and apparel. Comparing a supplements storefront against a book reseller on feedback count alone compares two different conversion regimes, and the list mixes them freely.
A company is not a storefront. The list ranks selling accounts. On 2026-09-06 the top 100 held “Pattern.” at rank 1 and “Pattern Products” at rank 10; “Ekouaer” at 25 and “Ekouaer loungewear” at 46; and four ThriftBooks storefronts at ranks 21, 33, 43 and 100. Whether you treat those as one seller or as seven changes the shape of the leaderboard. Nothing on the page aggregates them for you.
Not everything in the list is an independent third-party seller. Rank 11 on 2026-09-06 was Woot, which states on its own FAQ that “Woot! is an independent subsidiary of Amazon” (checked 2026-09-06). Amazon’s own retail business — the first-party half of the marketplace — is not in the list at all, because it does not collect seller feedback as a third-party storefront. A “top marketplace sellers” list is a top-3P-storefronts list with some Amazon-owned storefronts mixed in.
There is a fifth pressure worth naming plainly: feedback can be manipulated. The US Federal Trade Commission’s Trade Regulation Rule on the Use of Consumer Reviews and Testimonials was published at 89 FR 68034 on 2024-08-22 and took effect 2024-10-21 (Federal Register record, checked 2026-09-06). The FTC’s announcement of the final rule, dated August 14, 2024, says it “addresses reviews and testimonials that misrepresent that they are by someone who does not exist, such as AI-generated fake reviews” and “prohibits businesses from providing compensation or other incentives conditioned on the writing of consumer reviews expressing a particular sentiment, either positive or negative.” A rule exists because the behavior exists. If you want to pressure-test a specific storefront’s review profile, our roundup of Amazon fake review checker tools covers what those tools can and cannot detect.
Two things the page does not advertise
Both are properties of the live page rather than documented features of it, and both are easy for you to reproduce. The readings below are what the page returned on 2026-09-06.
The locked revenue figures change on every page load. Repeated fetches of the same Amazon.com list URL return the review columns unchanged while the locked Yearly Revenue cell returns a different number each time. Across repeated loads of the page on 2026-09-06, the Pattern. row held at 5,438 thirty-day reviews and 517,349 lifetime reviews throughout, while the Yearly Revenue cell on that same row came back as 704,307, 318,318, 838,666 and 627,352 on successive loads the same day. Those are placeholder digits rendered behind the lock, not a number that reloaded differently. The practical consequence: any article quoting a revenue figure “from the Marketplace Pulse top sellers page” without a subscription is quoting scrambled filler. If you see one, that is your tell.
The published rank is not a plain sort of the displayed 30-day column. In the Amazon.com list as served on 2026-09-06, 32 of the 100 rows showed a higher 30 Day Reviews value than the row directly above them. Rank 11 (Woot, 1,883) sat below rank 10 (Pattern Products, 1,810). Rank 29 (Filterbuy, 1,411) sat below rank 28 (My Great Goods, 1,001). The cause is not something the page discloses — a smoothed or lagged ranking window would produce exactly this, and so would several other designs. What is observable is that the visible column does not reconstruct the visible order, so treating a one-place gap as meaningful is unsupported by anything the page shows.
A checklist for reading any biggest-seller list
Run these six questions against any ranking of top Amazon marketplace sellers before you quote a position from it.
- What is the unit being counted? Feedback events, product reviews, estimated units, or modelled dollars. If the page does not say in one sentence, stop there.
- What is the window? A 30-day count is a snapshot. Q4 and Prime Day snapshots do not describe an average month, and a list built on a trailing year describes neither.
- Is the ranked entity a storefront or a company? Check for repeated brand stems before you say “the number one seller.”
- Is the dollar column free or paid? If it is paid, assume any figure reproduced elsewhere is unverified. Test it by reloading the page.
- Which marketplace? The Amazon.com list and the Amazon.co.uk list are separate tables with separate populations. A rank is only a rank inside one of them.
- Does the number have a source and a date on it? Marketplace Pulse dates its list on the page. Most articles that repost it do not carry that date forward, so a figure you find in a blog post may be many months stale.
That last one is the most common failure by a wide margin. A ranking is a dated observation, and a dated observation that lost its date has lost most of its value.
What to use when you need dollars
Feedback rankings answer “who moves a lot of orders right now.” They do not answer “who is big.” For size questions, go to figures that are either published by Amazon or explicitly modelled and labelled as such:
- Amazon’s own quarterly disclosure of third-party seller units as a share of paid units. Marketplace Pulse’s compiled series of that reported figure puts the last reported quarter, 2026 Q1, at 60.0% of worldwide units, down from 61.0% a year earlier (checked 2026-09-06). It is a real reported figure — but it is marketplace-wide, not per seller.
- Marketplace Pulse’s own research articles, which do publish dollar estimates with methodology attached, clearly separated from the free ranking table. Our guide to how many sellers are on Amazon walks through several of those estimates and who counted what.
- Category-level analysis rather than seller-level, when the question is really about demand — covered in most selling products on Amazon, and in the Amazon BSR guide for rank-based reading of individual products.
The pattern across all four of those pages is the same one this article argues for: name the metric, name the source, name the date, and never let a proxy get promoted to a measurement.
Frequently Asked Questions
Does Marketplace Pulse rank Amazon sellers by revenue?
No. The Amazon.com page states that “Seller rank is based on the number of reviews each seller received during the last 30 days” (checked 2026-09-06). A Yearly Revenue column exists on the page but is a locked, paid field.
Are the review counts the same as product reviews?
No. They are seller feedback — ratings left about the transaction and the storefront, not about an ASIN. Product reviews and seller feedback are separate systems with separate submission paths and separate rules.
How accurate is review count as a sales proxy?
Unquantified, by the source’s own account. Marketplace Pulse says the share of customers leaving seller feedback runs “anywhere in the range of 1–5%, depending on the product’s pricing and category” (checked 2026-09-06). Within one category the ordering is likely informative; across categories and across price points it is much weaker.
Who was the top Amazon marketplace seller in September 2026?
On the list dated September 5 and retrieved 2026-09-06, the top-ranked Amazon.com storefront was Pattern., with 5,438 seller reviews in the trailing 30 days. That is a feedback ranking for that window, not a statement that Pattern. sold the most.
Can I see the full top 100 for free?
Yes — the ranking table with rank, seller, country, 30-day reviews and lifetime reviews is public. The revenue column and the list download are the paid parts.
Conclusion
The public ranking of top Amazon marketplace sellers is a good dataset used badly. Marketplace Pulse is transparent about its basis in a sentence at the top of the page; almost everything downstream of it quietly upgrades “30-day feedback count” into “sales” and drops the date. The fix is not to distrust the list — it is to carry its label with it. When you cite a position from it, cite the metric, the marketplace, and the day you looked, and you will be more accurate than most of what is written about it.
Everything in this article was checked against the live pages named on 2026-09-06. Rankings change; the methodology sentence is the durable part.