Benchmark browse category level is a dimension in Amazon Ads unified reporting that records how deep in Amazon's product category tree the category used for a benchmark comparison sits. It exists because a benchmark is only as meaningful as the peer group behind it. A click-through rate compared against every brand in "Home & Kitchen" says something different from one compared against brands selling only pots and pans, and the level field tells the reader which of those two comparisons a row of data represents.

The field is one of seven in the Benchmark category of Amazon's unified reporting dimensions reference, updated on November 11, 2026. Unlike most unified dimensions, it carries no "former name" from the legacy sponsored ads or Amazon DSP reports.

Where the field sits

Amazon's unified reporting system, generally available in the Ads Console since June 8, 2026, organises its 156 dimension attributes into 11 categories: Currency, Time, Level of detail, Product, Supply, Technology, Geography, Audience, Targeting, Conversion and Benchmark. The Benchmark block contains:

  • Benchmark brand ID and Benchmark brand name
  • Benchmark browse category ID
  • Benchmark browse category path
  • Benchmark browse category level
  • Benchmark browse category marketplace
  • Benchmark campaign goal

Amazon's reference says the first six are "derived from the products promoted in the campaign", while the campaign goal field is "derived from campaign goal and optimization settings", according to PPC Land's reading of the page. All seven carry markers for DVA+, Sponsored Products, Sponsored Brands and Sponsored Display. The four browse category fields describe one object from four angles: the ID is the category's numeric identifier, the path is its readable chain of names, the marketplace is the Amazon store whose tree it belongs to, and the level is its depth in that tree.

None of these fields is a metric. Amazon's Support Center page on benchmark metrics describes them as showing "the peer group for each comparison", according to PPC Land's November 12 analysis.

How the browse tree works

The level only makes sense against the structure it measures. Amazon files every product in a hierarchy of browse nodes. According to Amazon's Creators API documentation, "a BrowseNodeId is a unique ID assigned by Amazon that identifies a product category/sub-category." Each node has ancestors running up to a root node, which has none, and children running down to leaf nodes, which have none.

Amazon's documentation gives a worked example. Node 3040, "Mexico", sits under "Explore the World" (3023), "Geography & Cultures" (3344091011), "Children's Books" (4), "Subjects" (1000) and finally "Books" (283155), the root. Display names can mislead without that ancestry: node 3060 reads simply "Orphans & Foster Homes".

Each marketplace runs its own tree. A vendor guide from WISEPIM, updated October 11, 2026, states that a node ID valid on Amazon.com "may not exist or may point to a different category on Amazon.de", and puts the maximum depth at eight levels; Amazon's API reference gives no limit. Hence the marketplace field: a category ID means nothing without its store.

What the level value tells the reader

The level field reduces the path to a single number: how many steps the benchmark category sits from the top of the tree. A low level indicates a broad department; a high level indicates a narrow sub-category. In the books example above, "Books" is the root and "Mexico" sits five steps beneath it.

Amazon's dimensions reference does not state, in the material reviewed for this article, whether the root counts as level 0 or level 1. Readers comparing levels across exports should check the convention against the path field, which spells out each step and so allows the depth to be counted directly.

The practical use is filtering and sanity checking. An analyst with hundreds of campaigns can group benchmark rows by level to see which comparisons rest on broad peer pools and which on narrow ones, without parsing text paths.

How benchmark rows are populated

The browse category is not chosen by the advertiser. According to the reference, it is derived from the products promoted in the campaign: the advertised items map to browse nodes, and the benchmark category follows from them. Amazon has not published the rule used when a campaign promotes products from several nodes, or when a product sits in more than one node.

Peer selection then happens within that category. Amazon's benchmarks white paper, "The New Standard: How Benchmarks Are Transforming Performance Evaluation", promoted on March 27, 2026, describes a system called PeerSets that "identifies direct competitor brands for any given brand within a category context". It combines a K-Nearest Neighbors (KNN) statistical model, which measures "the mathematical distance between brands based on their brand-level revenue and average sales price within a category", with a large language model (LLM) that evaluates "direct competition dynamics, market segment overlap, and customer base similarities".

Campaigns must pass activity and attribute thresholds for which the paper gives no numbers, and two privacy rules apply. The paper requires "a minimum sample size of at least five peer brands to construct any benchmark", and caps "any single brand's contribution at no more than 49% of the total" spend weight.

The five-brand floor has a direct bearing on the level field. The deeper the category, the fewer brands it holds. Where a narrow node cannot supply five comparable peers, "no benchmark is surfaced", according to the paper. Whether the system then moves up the tree to a broader level, or simply shows nothing, is not documented.

The values that appear alongside the level are the 25th, 50th and 75th percentiles of eight metrics, among them click-through rate (CTR), cost per click (CPC), cost per thousand impressions (CPM) and three new-to-brand purchase measures. Unified reporting labels them in columns such as "CTR (p25)", "CTR (p50)" and "CTR (p75)".

Origin and evolution

Amazon introduced self-service benchmarks on November 18, 2025, as two report types in its Reporting API: dspBenchmarks for Amazon DSP and crossProgramBenchmarks for sponsored ads. The beta was limited to the United States, and Amazon's launch page promised "transparent methodology" without publishing one.

The white paper followed on March 27, 2026, supplying the KNN, LLM, five-brand and 49% rules. On May 18, 2026, benchmarks reached general availability in 18 markets, from the United States and Brazil to Saudi Arabia, the United Arab Emirates and Egypt.

The dimension itself arrived with the move into unified reporting. The November 11, 2026 dimensions reference is the first public listing of the seven Benchmark fields, which PPC Land described as "the newest block, with no former name columns at all".

Why it matters

Benchmarks change how campaign performance is judged. A CPC of $1.20 is neither high nor low until it is compared with something, and Amazon now supplies that comparison inside the same report as the campaign's own numbers. The browse category fields define what the comparison is against. Without them, a p50 value is a number with no stated population.

The question has commercial weight: PPC Land found that none of Amazon's 16 recommended actions attached to benchmark metrics suggests cutting spend. A broad-category benchmark can make a narrow product look weak or strong for reasons unrelated to its advertising, and the level field is the quickest indicator of how broad the comparison was.

Limitations and disputes

The methodology is Amazon's own and cannot be audited. The peer brands are not named, the LLM component is described only in general terms, and the activity thresholds carry no figures. A brand can see the category level of its peer set but not the brands in it.

Coverage statements conflict. Amazon's May 2026 general availability page lists Amazon DSP, Sponsored Products, Sponsored Brands, Sponsored Display and Sponsored TV. The November 11, 2026 Support Center page on benchmark metrics says metrics in unified reporting "are available for Sponsored Products, Sponsored Brands, Streaming TV, and DVA+", omitting Sponsored Display. The dimensions reference, meanwhile, marks all seven Benchmark fields for Sponsored Display, and says of Sponsored TV: "We do not support Streaming TV Smart mode/ Sponsored TV yet."

Migration is incomplete. Amazon's migration page, updated November 12, 2026, lists the benchmark report among ten report types not yet available in unified reporting, even though benchmark columns and dimensions already exist there. Legacy report pages become read-only on December 17, 2026 and are scheduled to close on January 31, 2027.

Benchmark browse category level is not...

Benchmark browse category path. The path is the readable chain of category names from root to node. The level is the count of steps in that chain. Both describe the same category.

Advertised product category and subcategory. These unified dimensions describe the category of the product an advertisement promoted, row by row. The benchmark fields describe the category used to build the peer comparison, which is derived from the campaign's products but is a separate attribute.

Category targeting. In sponsored ads and DVA, advertisers can target products or shoppers in particular categories. That is a buying setting. The benchmark level is a reporting label and changes nothing about delivery.

Best Sellers Rank category. The category shown next to a product's sales rank on its detail page is a shopper-facing display. It may share a node with the benchmark category but is not the same field.

Recent developments

Most of the public detail arrived this week. On November 11, 2026, Amazon listed the Benchmark block in its dimensions reference for the first time and updated its Support Center page on percentiles, peer groups and recommended actions. Once legacy pages close on January 31, 2027, unified reporting will be the only route to these fields.

Timeline

  • November 11, 2025 - Unified reporting opens in beta.
  • November 18, 2025 - Amazon introduces self-service benchmarks in the United States through dspBenchmarks and crossProgramBenchmarks report types.
  • March 27, 2026 - Amazon promotes its benchmarks white paper, describing PeerSets, the five-brand minimum and the 49% weight cap.
  • May 18, 2026 - Benchmarks reach general availability in 18 markets with eight metrics.
  • June 8, 2026 - Unified reporting reaches general availability in the Ads Console.
  • September 29, 2026 - Amazon DSP renamed DVA.
  • November 11, 2026 - Amazon's dimensions reference lists seven Benchmark fields, including Benchmark browse category level; Support Center benchmark metrics page updated.
  • November 12, 2026 - Migration page lists the benchmark report among ten types not yet in unified reporting.
  • December 17, 2026 - Legacy report pages scheduled to become read-only.
  • January 31, 2027 - Legacy report pages scheduled for decommissioning.

Summary

Who: Amazon Ads defines and populates the field; advertisers, agencies and reporting tools using benchmarks for DVA+, Sponsored Products, Sponsored Brands and Sponsored Display read it.

What: A unified reporting dimension giving the depth in Amazon's browse tree of the category used to build a benchmark peer group, alongside that category's ID, path and marketplace.

When: Benchmarks launched on November 18, 2025 and reached general availability on May 18, 2026; the field was first listed in Amazon's dimensions reference on November 11, 2026.

Where: In Amazon Ads unified reporting in the Ads Console and the Reporting API, across the 18 benchmark markets, with values tied to each marketplace's own browse tree.

Why: A benchmark is only interpretable if its peer group is known. The level shows whether a campaign is being compared with a broad department or a narrow sub-category, which matters because deeper categories hold fewer brands and Amazon withholds benchmarks with fewer than five peers.