Results: Amazon Brazil, overall cut
177 products from 15 brands, weeks from Sep 21, 2025 to Sep 5, 2026, cutoff date Sep 24, 2026. In each row, half of the products fall between the 25th and the 75th percentile; the median is the middle product.
| Rate | 25th | Median | 75th |
|---|---|---|---|
| Click-through (CTR) | 1.5% | 2.1% | 3.2% |
| Add to cart | 14.6% | 22.7% | 32.7% |
| Close | 9.1% | 14.3% | 25.0% |
| Search conversion | 1.5% | 3.2% | 6.0% |
How to read it: if, out of every 100 clicks in search, your product sells 2, it is below the median (3.2) and above the 25th percentile (1.5). If the click is good and the conversion isn't, the problem usually lies after the click — the product page, price or shipping. Category cuts come in when each category reaches 10 brands and 100 products.
Why the close rate looks low: Amazon counts every add to cart, not every buyer — the same buyer can add more than once, and the purchase can happen in another week or come from another search. Compare your close rate against this yardstick, not against the cart rates of an own-brand store.
Why a benchmark
"Is my conversion good?" is the most common question from people who sell on Amazon — and the least answered. A conversion rate alone says nothing: 8% is great in a category where everyone converts 4% and poor in one where everyone converts 15%. No source publishes, for Amazon Brazil, the typical conversion by funnel stage and by category. This benchmark exists for that.
The methodology below was published before the numbers were final, on purpose: anyone using the data should be able to know exactly how it was calculated and what it doesn't show.
What the benchmark measures
Four rates, one for each stage of the buyer's path in an Amazon search:
| Rate | Calculation | What it answers |
|---|---|---|
| Click-through (CTR) | clicks ÷ impressions | Did people who saw the product in search click? |
| Add to cart | add-to-carts ÷ clicks | Did people who opened the page want to buy? |
| Close | purchases ÷ add-to-carts | Did people who added it to the cart finish the purchase? |
| Search conversion | purchases ÷ clicks | Out of every click, how many became a purchase? |
For each rate we publish the median and the quartiles (25th and 75th): half of the products are above the median and half below; between the quartiles sits the middle half. We use the median, not the average, because a few products with extreme numbers pull the average away from the typical product.
Where the data comes from
From Amazon Brand Analytics' Search Query Performance report: weekly (Sunday-to-Saturday weeks), by product and by search, with impressions, clicks, add-to-carts and purchases. Amazon delivers this report only to brand owners in Brand Registry, and only for the searches where the brand appeared.
The sample is the Amazon Brazil brand accounts connected to EcomVision. The numbers are used only in aggregate: the unit of analysis is the product in a week, and what is published is the distribution of many products from many brands.
How it's calculated
- Period: up to 52 closed weeks at the cutoff date, stated at publication. The 2 most recent weeks are left out, because Amazon may still reprocess them. In this edition, collection began in September 2025 and the period covers Sep 21, 2025 to Sep 5, 2026.
- Product: variations (color, size, flavor) are summed into the parent product, the way the buyer sees the product.
- Minimum volume per product: a product with fewer than 20 clicks in the period is excluded — with little volume, the rate becomes luck.
- Rate per product: we sum the product's impressions, clicks, carts and purchases over the period and calculate the four rates from the sums, not from the average of the weeks.
- Distribution: the median and quartiles come from the set of products in each cut.
Published cuts
The overall Amazon Brazil benchmark and one cut per Amazon main category. A cut is only published if it meets all the rules below:
- at least 10 different brands;
- at least 100 products with 20 clicks or more in the period;
Cap per brand: no brand weighs more than 20% of a cut's products. When a brand goes over that, only its products with the most clicks are included, up to the cap — the rest is left out of that cut. That way the median describes the market, not the brand with the most products in the sample.
Each published cut carries its N — how many brands and how many products went in. A category that doesn't reach the minimum doesn't appear, instead of appearing with a number that would represent a few accounts.
What is never published
No ASIN, brand, store, seller or account — not in a table, not in a chart, not in an example. The 20% cap per brand exists precisely so that no cut's number is, in practice, the number of a single brand.
What this benchmark doesn't show
- It isn't all of Amazon. The sample is brands with Brand Registry connected to EcomVision. Sellers without a registered brand aren't included, and the sample may have more professional operations than the marketplace average.
- It's search conversion. The report covers purchases that came from searches, not all sales — external traffic, ads outside search and direct access are left out.
- Cart is aggregate. Amazon reports how many times the product was added to the cart, not each buyer's cart. Close and abandonment are aggregate readings.
- Amazon may revise weeks. Closed weeks are sometimes reprocessed; that's why only stable weeks are included at the cutoff date.
How to compare your account in the meantime
Search Query Performance itself brings, next to your numbers, the total for all sellers in the same searches. Comparing each stage against that total — impression, click, cart, purchase — is the fairest way to know where your product loses the buyer, because the yardstick is the same search. That's what EcomVision's Shopper Funnel does, with an index per stage (1.00 = same as the search).
Schedule
Methodology and first numbers published on Sep 24, 2026, with the overall Amazon Brazil benchmark. Category cuts come in as each category reaches 10 brands and 100 products. Updated annually, always at this URL.