October 26, 2026

Amazon Attribution: How Attribution Tags Measure Off-Amazon Traffic, and How to Read Them in Claude

What Amazon Attribution measures, how 14-day last-click tags work, why small rows show zeros, how Brand Referral Bonus fits, and how to ask Claude.

Amazon Attribution answers a question Sponsored Ads cannot: when you send traffic to Amazon from Google, Meta, TikTok, an email or a creator's link, what happens after the click? It measures detail page views, add-to-carts and purchases for traffic that started outside Amazon, and Amazon does not charge for it.

This guide explains how Attribution tags work, what the reports contain, the rules that make the numbers easy to misread, how the Brand Referral Bonus fits in, and how to ask Claude which off-Amazon channel actually sold units, using the Attribution data DataDoe collects from Amazon Ads.

What Amazon Attribution is

Attribution is a measurement console inside Amazon Ads for brand owners: sellers enrolled in Brand Registry, vendors, and the agencies working for them. You create a campaign and ad groups in the console, and Amazon gives you a tagged URL for each placement. You use that URL wherever the traffic starts: a search ad, a social post, a newsletter, an influencer's bio.

When a shopper clicks the tag and lands on Amazon, Amazon follows what they do next. It does not change how your ads run on Google or Meta. It only reports, on the Amazon side, what the traffic did.

How an Attribution tag measures a sale

  • One tag per placement. The tag carries the campaign, the ad group and the publisher, so a Meta carousel and a Google search ad get different tags even when they point to the same ASIN.
  • 14-day last click. A purchase counts for the last tagged click within 14 days before it. Earlier clicks on other tags get nothing. It is not multi-touch attribution.
  • Promoted and brand halo. A shopper who clicks a tag for one product and buys another product of your brand still counts. The reports split the two: purchases of the promoted ASIN, and "total" figures that include the halo.
  • New-to-brand. At product level, Amazon marks purchases from shoppers who had not bought from your brand before.

What the reports contain

In DataDoe, Attribution comes in three tables from your Amazon Ads connection:

  • Performance by ad group (amazon_ads_attribution_performance_by_ad_group_by_date): clicks, detail page views, add-to-carts, purchases, units and sales per day, campaign, ad group and publisher, for the promoted products and in total. This is the table for campaign totals, and the only one with the Brand Referral Bonus.
  • Performance by creative (amazon_ads_attribution_performance_by_creative_by_date): the same metrics one level down, per tagged link.
  • Products (amazon_ads_attribution_products_by_date): which ASINs the traffic bought, with brand halo and new-to-brand columns, for the top 100 products per campaign.

On the first load, the two performance tables reach back about 13 months and the products table about 90 days. After that, the last 30 days are re-read every day, because Amazon keeps updating purchases inside the 14-day window. None of the three is part of the default dataset, so check that they are enabled in Settings > Data.

For books, the tables also carry Kindle Edition Normalized Pages read and the estimated royalties from them.

Three rules that keep the numbers honest

  1. Small rows are masked. Amazon reports zero for every metric on a row with fewer than 10 lifetime clicks. A zero on a new or small tag means "hidden", not "nothing happened".
  2. Don't sum creatives into ad groups. Because of the masking, creative rows will not add up to the ad group total. Take totals from the ad group table.
  3. Don't add Attribution to Sponsored Ads. Attribution measures traffic that started off Amazon, with its own window. Sponsored Ads have their own conversion windows for clicks on Amazon. A sale should be read in one report or the other, not added from both.

Brand Referral Bonus, read from the same data

The Brand Referral Bonus (BRB) is Amazon's credit for brand-registered sellers who drive outside traffic to Amazon. You enroll, use Attribution tags on your external marketing, and Amazon pays a bonus on qualifying sales as a credit against your referral fees.

In DataDoe, brb_bonus_amount on the ad group table holds Amazon's estimated, ad-attributed bonus in local currency. It is filled only for enrolled sellers and only at ad group level. That makes it easy to put the bonus next to the channel that earned it: "Which publishers earned the most Brand Referral Bonus last quarter?" A separate post on BRB follows later in this series.

Ask Claude which channel actually sold

With DataDoe connected to Claude, the Attribution console's tabs become questions:

  • "Rank last month's off-Amazon publishers by attributed sales, with clicks, detail page views and purchases. Use the ad group table."
  • "For each campaign, what share of attributed sales came from brand halo rather than the promoted product?"
  • "Which ASINs did the TikTok traffic buy, and how many of those purchases were new-to-brand?"
  • "Compare click-to-purchase rate for the Google and Meta ad groups over the last 90 days. Flag rows that may be masked."
  • "Put attributed sales and Brand Referral Bonus side by side per publisher for Q3."

Claude can also put Attribution next to the rest of your account in the same answer, such as total sales of the promoted ASINs from the profit tables, so you can see how much of a product's demand started outside Amazon.

Return on spend needs your spend

Attribution knows what happened on Amazon, not what you paid the publisher. DataDoe does not read Google Ads or Meta spend either. To get return on ad spend per channel, give Claude the spend figures from those platforms, and it will divide attributed sales by them. Keep in mind that the sales side is still 14-day last click on Amazon only.

What Attribution in DataDoe is not

  • It does not create tags or campaigns. There are no Attribution Actions; you create tags in the Amazon Ads console.
  • It is not multi-touch. Every sale belongs to the last tagged click.
  • It does not show off-Amazon spend, impressions or ROAS from the ad platforms themselves.
  • It cannot unmask small rows. Amazon hides them before the data reaches anyone.

For everything else Claude can read in your ad account, see Amazon Ads in Claude and ChatGPT. For how on-Amazon ad sales are judged against total sales, see ACoS vs TACoS. For the Brand Registry data that goes with it, see Amazon Brand Registry data.

Questions people ask

What is Amazon Attribution?

A free Amazon Ads tool that measures what shoppers do on Amazon after clicking a tagged link outside Amazon: detail page views, add-to-carts, purchases and sales.

What is the Amazon Attribution window?

14 days, last click. A purchase is credited to the last tagged click in the 14 days before it.

Why are some Attribution rows all zeros?

Amazon masks the metrics of any row with fewer than 10 lifetime clicks. The zeros mean hidden, not zero sales.

Can Claude create Amazon Attribution tags?

No. DataDoe reads Attribution reports, and tags are created in the Amazon Ads console.

How do I see my Brand Referral Bonus by channel?

If you are enrolled, DataDoe's ad group table carries Amazon's estimated bonus per ad group and publisher, so Claude can group it by channel.

• Get started

Put this to work on your own Amazon data.

Connect Seller Central, Vendor Central and Amazon Ads in about 5 minutes, then ask your AI about your own numbers instead of the examples in this post.

  • Seller, Vendor and Ads
  • Any AI client
  • An account manager for every customer
  • $97/mo, 14-day free trial