Amazon Attribution in 2026: Measuring External Traffic

Quick answer: Amazon Attribution is Amazon's free measurement system for advertisers sending traffic from non-Amazon channels. Attribution tags connect an external click with later Amazon shopping activity under Amazon's reporting rules. The resulting purchases are attributed conversions, not proof the external campaign caused every sale. Evaluate attributed revenue, contribution margin, and likely incrementality together, not the headline revenue number alone.

A skincare seller paid three creators $8,000 to promote a launch. Amazon Attribution reported $29,400 in purchases, and the marketing manager called it a success. The finance manager asked a different question: how many of those customers would have bought anyway? During the campaign, the brand also ran Sponsored Products, emailed existing customers, and offered a coupon. One creator's link was copied into a deal group, another creator linked to the brand Store rather than the exact ASIN, and brand-halo purchases appeared for related products. The campaign had clearly generated activity. The exact amount of incremental profit was much less obvious.

What an attribution tag does

Amazon Attribution campaigns generate tags applied to external marketing links. The URL parameters let Amazon associate later shopping actions with the external interaction under its attribution methodology. Reported events can include detail page views, add-to-cart actions, and purchases, and depending on account type, sellers may also see brand-halo activity where a shopper interacts with the promoted product but purchases another product from the brand. The tag is a measurement link, not a coupon code, and it doesn't identify a customer to the seller.

Why Amazon Attribution and Seller Central do not match

They answer different questions. Seller Central reports total Amazon orders under its own report definitions. Attribution reports shopping activity associated with tagged external interactions specifically. Differences come from attribution windows, click timing versus order timing, cancellations and returns, brand-halo purchases, duplicate customer journeys across channels, and links that were copied or stripped of parameters. A mismatch is expected, the task is understanding whether the difference is reasonable and consistent over time.

The 2026 attribution-model change sellers should note

Amazon Ads announced a shopping-signal-enhanced last-touch attribution model for Amazon Store ads beginning January 1, 2026. The exact impact depends on the ad product and report being used. Before publishing a benchmark or comparing year-over-year performance, record the report name, ad product, attribution model, and conversion window used, since a model change can make a campaign appear better or worse without customer behavior actually changing to the same degree.

Build the campaign structure before generating links

A common mistake is creating one tag for an entire channel, which answers whether Instagram sent activity but not which creator, placement, or creative did the work. A useful hierarchy separates channel, publisher, campaign, ad group, and creative, with stable naming rules, if one team member writes IG-CreatorA and another writes instagram_creator_a, reporting becomes harder to maintain over time.

The profitability calculation

Start with attributed sales, but don't stop there. Attributed contribution equals attributed revenue minus Amazon fees, product cost, expected returns, coupon cost, and creator or media spend. For the creator campaign: $29,400 attributed revenue, minus $9,100 Amazon and fulfillment fees, $6,300 product and inbound cost, $1,500 expected returns, $2,050 coupon cost, and $8,000 creator fees, leaves an estimated contribution of $2,450. The campaign is profitable on an attributed basis, but the margin is much smaller than the revenue headline suggested.

Attributed profit is not incremental profit

Incrementality asks what happened because of the campaign that wouldn't otherwise have happened. Attribution doesn't automatically answer that, a loyal customer might click a creator link and buy a product they planned to reorder anyway. Estimated incremental contribution equals attributed contribution times an estimated incremental share, if the team believes 50% to 80% of the attributed contribution is incremental, the range in the example above is $1,225 to $1,960. Justify that estimate using new-to-brand indicators, branded-search movement, geographic or time-based holdouts, and repeat customer share, not a weak assumption dressed up with two decimal places.

Brand halo: valuable, but easy to misuse

A campaign promoting Product A generating purchases of Products B and C can be real value, but it complicates margin analysis since the purchased products may carry very different contribution rates. Separate promoted-product purchases from brand-halo purchases, and flag products with high return rates or thin margin that absorbed the traffic. A campaign can look strong in revenue because customers bought a high-priced related item while the promoted product itself failed to convert.

Case study: the link that escaped the intended audience

One creator's tagged URL was reposted in a public deals forum, and attribution purchases surged for two days. The marketing team initially credited the creator's content. The order profile told a different story: coupon redemptions rose sharply, average order value fell, repeat-customer share increased, and the creator's own video views didn't spike at all. The traffic was still connected to the tag, but the audience and reason for purchase had changed entirely. The team separated the period in its analysis and stopped using the result as a creator benchmark.

Edge cases

A tag leading to a Store rather than an ASIN supports discovery and halo sales but makes the path less direct, use product-level tags when the creative is about one product. A shopper clicking several tagged links means the reporting model decides which interaction receives credit, don't add campaign reports together and assume every line represents a unique order. Purchases appearing after the campaign ends reflect the conversion window, keep reporting open long enough before declaring final results. And external traffic can damage conversion if poorly targeted, track the ASIN's total conversion and customer questions, not only attributed revenue.

A simple incrementality test for smaller brands

A perfect controlled study is rarely available to a small seller, but a basic holdout beats no counterfactual at all. Choose comparable days or audience segments, keep price and Amazon advertising as stable as possible, and run the external campaign in the test group while observing the difference from baseline. If Attribution reports 300 purchases but total sales rise by only 40 above a stable baseline, the campaign probably didn't create 300 fully incremental orders, though it may still have value through customer acquisition or future repeat demand.

---

Amazon Attribution remains the source for tagged off-Amazon interactions. EcomSanity can help show what happened to the wider Amazon business: total ASIN sales during the campaign, inventory depletion, Buy Box changes, and returns after the campaign, useful when a campaign looks profitable in isolation but causes a stockout or shifts sales away from full-price traffic.