Original study · 5 October 2026 · Chilview LLC
Who Really Holds Amazon's Buy Box? 57,410 Observed Visits Say Third-Party Sellers Hold It 92.8% of the Time
Summary: We analyzed 57,410 real Amazon product-page visits captured by Chilview between 12 May 2026 and 5 October 2026. Amazon held the Buy Box on 7.2% of visits; third-party sellers held it on 92.8% (92.0% fulfilled by Amazon, 0.8% merchant-fulfilled). These are observed holder shares on captured visits — not units sold, and not a designed category panel.
Author: Chilview LLC · Date published: 2026-10-05
Key findings
1. Third-party sellers dominate the Buy Box on observed visits.
Out of 57,410 US visits across 49,405 distinct products, the Buy Box winner was Amazon on 4,124 visits (7.18%), a third-party FBA seller on 52,806 visits (91.98%), and a third-party merchant-fulfilled seller on 480 visits (0.84%).
| Winner type | Visits | Share |
|---|---|---|
| Amazon | 4,124 | 7.18% |
| Third-party FBA | 52,806 | 91.98% |
| Third-party merchant | 480 | 0.84% |
| Total | 57,410 | 100% |
2. The often-quoted "62%" measures something different.
Amazon reported that third-party sellers accounted for 62% of units sold in Q4 2024 (via Marketplace Pulse). That is a units-sold share. Our 92.8% is an observed holder share on captured product-page visits. The two numbers answer different questions — units sold tells you volume mix, holder share tells you what a shopper actually sees on the page. Both are true; neither replaces the other.
3. Buy Box holder type rarely flips between visits.
Among the 5,456 products visited more than once (8,005 successive visit pairs), the winner type changed 302 times (3.8% of pairs). 251 of the revisited products (4.6%) ever changed winner type. Note: most products in this dataset were seen only once (44,949 of 49,405), so this measures type stability on revisited listings, not full rotation.
4. Listings with 3+ sellers are a small minority of captures.
Only ~18 US products in the dataset had a stored FBA seller count of 3 or more, and 6 had 3+ wholesale offers counted. No Electronics product qualified. (A handful of rows from 12 June 2026 showed FBA counts in the hundreds — these look like parse errors and were excluded as unverified.)
Methodology
- Source: Chilview's own product-data API, US marketplace.
- What was analyzed: the
buy_box_historytable — one row per product-page visit where the extension captured the Buy Box panel. - Sample: 57,410 visits across 49,405 distinct ASINs, captured 12 May 2026 – 5 October 2026. 56,840 of the 57,410 visits (99%) occurred in the last 30 days of the window.
- How captures happen: a row is written when a Chilview Founder opens a product page with the extension installed. Chilview does not crawl Amazon on a schedule — these are visit captures only. This makes the sample a convenience sample of products Founders actually opened, not a designed or random sample of Amazon's catalog.
- What was recorded per visit: winner type (amazon / third_party_fba / third_party_merchant), price, and timestamp. No seller names were collected or used — Chilview collects no personal info, only product information.
- Cleaning: QA/test ASINs excluded. Rows with implausible FBA counts (hundreds, clustered 12 June 2026) treated as parse errors and excluded from seller-depth claims.
- What this is NOT: because the median product was visited once, this dataset cannot measure how often the Buy Box rotates on a single listing over time. The flip figures above cover only the 5,456 revisited products.
FAQ
Frequently asked questions
Does this mean third-party sellers get 92.8% of Amazon's sales?
No. Our number is the share of observed Buy Box holdings on captured visits. Amazon's reported 62% third-party unit share (Q4 2024) is the sales-volume figure. Holder share tells you what shoppers see; unit share tells you who ships the volume.
How is this data collected?
When a Chilview user opens an Amazon product page, the extension reads the publicly visible Buy Box panel (winner type and price) and stores it. No crawling, no Seller Central, no personal data — only product information from pages the user already opened.
Why don't the numbers match the "62%" I keep seeing?
Because they measure different things. Anyone quoting 62% as a Buy Box holder share is misusing the statistic.
Can the Buy Box change between two visits to the same page?
Yes. On the 5,456 products we saw more than once, the winner type changed on 3.8% of successive visit pairs. The holder can rotate as often as every few minutes on competitive listings, so any single visit is a snapshot, not a permanent state.
Which categories were studied?
This is a visit-weighted sample, not a category-balanced panel — it reflects whichever products Founders opened. Top-level department is not stored, so no category-level claims are made.
Will you update this study?
Yes — we plan to refresh these figures periodically as new visits are captured. Each update will carry its own date range.
Next
Keep researching with Chilview
Chilview is a privacy-first Amazon product research Chrome extension for Chrome and Brave. $30 one-time for the first 500 Founder seats; then $10/month. 3-day trial, no credit card.