Amazon DSP is a programmatic advertising platform that enables automated, data-driven ad placements using Amazon’s audience insights. Unlike Amazon’s Sponsored Products or Sponsored Brands (which show ads on Amazon’s own site/app), DSP campaigns can reach users across the wider internet, including mobile apps and even streaming TV, using Amazon targeting signals. This means an advertiser can, for example, retarget someone who viewed a product on Amazon by showing them a banner ad on a news site or a video ad on a streaming service. Notably, Amazon DSP isn’t limited to Amazon sellers – even brands that don’t sell on Amazon can use it to tap into Amazon’s audience data for their marketing. However, Amazon DSP is generally positioned as a premium solution, so it’s most effective for advertisers with sufficient budget and programmatic expertise.
Why focus on measurement? In programmatic campaigns like those run on DSP, performance isn’t as straightforward as pay-per-click ads. Ads may serve at various points in the customer journey (awareness, consideration, retargeting), and conversions might happen days after exposure. For seasoned advertisers, accurately attributing outcomes to the DSP is essential to justify the investment and refine the strategy. Next, we’ll examine how Amazon DSP approaches attribution.
Robust measurement and attribution are the backbone of successful Amazon DSP campaigns. Here are key reasons why attribution matters for advertisers:
In summary, without solid measurement, you’re flying blind with your DSP campaigns. Now let’s look at how Amazon DSP actually attributes conversions to give you those insights.
Default attribution model: Amazon DSP uses a 14-day last-touch attribution model by default. This means that when a conversion (e.g. a purchase) occurs, the DSP will credit it to the last ad interaction (impression or click) that happened within the 14 days prior. In practice, if a customer clicks a DSP ad and buys the product a week later, that sale will be attributed to the DSP campaign. If they only viewed an ad (without clicking) and later purchased, that can also be attributed as a view-through conversion under the last-touch logic.
Click vs. view attribution window: For Amazon DSP, the attribution windows typically differ for click-through vs. view-through events. A common setup is a 14-day window for clicks and a longer 30-day window for view-through conversions. In other words, if someone sees your DSP ad but doesn’t click, then purchases your product within 30 days, the DSP may count that as a conversion from a viewed ad. If they clicked the ad, the conversion must happen within 14 days to count. This extended view-through window helps capture the impact of those upper-funnel impressions that influence buyers over a longer consideration cycle. (By contrast, Amazon’s own Sponsored Product ads often use a much shorter 7-day or even 24-hour attribution window in some cases, so DSP’s longer windows are a significant advantage in tracking delayed conversions.)
New-to-brand metrics: Amazon DSP reporting includes New-to-Brand (NTB) metrics that indicate how many conversions are coming from customers who have never bought from your brand before (or at least not in the past 12 months). This is crucial for brands focused on customer acquisition. A DSP campaign might show, for example, that 40% of its sales were “new-to-brand”, meaning the ads helped acquire brand-new customers. As mentioned, advertisers who use DSP in tandem with other ads often see a notable uptick in new-to-brand purchases – evidence that DSP can expand reach beyond your existing customer base.
Amazon’s evolving attribution: It’s worth noting that Amazon is continuously improving its attribution capabilities. While last-touch 14-day is the standard, Amazon has introduced modeled attribution and is piloting multi-touch models to give a more holistic view. In late 2024, Amazon DSP rolled out modeled conversion attribution for off-Amazon conversions, which uses machine learning to estimate and credit conversions that can’t be directly observed (more on this shortly)orange142.com. And starting in 2025, Amazon is rolling out a new Multi-Touch Attribution (MTA) system that distributes credit across multiple ad touchpoints, rather than only the last touch. The MTA approach leverages Amazon’s vast data (shopping, streaming, browsing signals) to assign proportional credit to each interaction in a 30-day path to conversion. The result for advertisers will be deeper insight into how upper-funnel ads (like streaming TV or Sponsored Brands video) contribute alongside lower-funnel ads to drive a sale. In short, attribution on Amazon DSP is moving from a single-touch view to a more full-funnel, multi-touch view.
For now, understanding the default 14-day last-touch model (with 30-day view-through) is key to reading your DSP reports. Next, we address one of the trickiest aspects of Amazon DSP measurement: tracking off-Amazon conversions.
One of Amazon DSP’s unique benefits is the ability to drive traffic outside of Amazon, such as to a brand’s own direct-to-consumer (D2C) website or a special landing page. For example, a skincare brand might use DSP to show ads that, when clicked, take shoppers directly to the brand’s official website to buy (instead of an Amazon product page). Measuring conversions in these cases – sales that happen off Amazon – comes with special challenges.
Amazon Pixel & Attribution tags: To track off-Amazon outcomes, Amazon provides tools like an Amazon tracking pixel (a snippet of code to place on your website) and Amazon Attribution tags for your DSP ads. When setting up a DSP line item that drives to an external URL, advertisers can generate an Amazon attribution tag (a unique parameter to append to the landing page URL) or use Amazon’s pixel. These allow Amazon to recognize when a DSP-driven user performs a conversion action on your site. For instance, if someone clicked a DSP ad to your website and makes a purchase there, the Amazon pixel on the purchase confirmation page would report that back, so the DSP campaign gets credit.
However, not all conversions will be traceable. There are cases where the pixel might not fire (if the user has certain privacy settings or if there were technical issues), or the user might convert through a different path later (e.g. sees the ad, later visits the site from a different device or via Google search). Amazon’s solution to this gap is the modeled attribution mentioned earlier – Amazon DSP can model some off-Amazon conversions based on probabilistic methods when direct tracking is missing. In August 2024, Amazon introduced this modeled attribution feature to better account for “view-through” impact on external sites, combining measured and modeled results into unified reporting. This helps give a more complete picture of campaign ROI across both Amazon and non-Amazon conversions.
Even with these improvements, advertisers are advised to look at aggregate lift in their own D2C analytics. Many brands will compare their overall D2C sales trends before and after running DSP ads to estimate the campaign’s impact, rather than relying solely on Amazon’s attributed numbers. This is because any attribution system may undercount some conversions. Savvy marketers often triangulate multiple data points – Amazon’s reports, Google Analytics, and backend sales data – to understand the full effect.
In summary, measuring off-Amazon outcomes requires using Amazon’s attribution tools (pixel/tag) and being comfortable with some estimation. Be prepared to supplement Amazon’s data with your own analytics to get the full story of how your DSP investment is paying off outside the Amazon ecosystem.
As Amazon’s advertising offerings mature, advanced measurement tools have emerged to help advertisers gain deeper insights from DSP and beyond. A key platform here is Amazon Marketing Cloud (AMC) – a privacy-safe data clean room where advertisers can run custom analytics across multiple Amazon ad products. Many sophisticated advertisers and agencies use AMC alongside DSP to perform attribution analyses that aren’t possible in standard reports. For example, you can query AMC to see the sequence of ad exposures that lead to conversion (path analysis) or to calculate your own multi-touch attribution model giving fractional credit to DSP and Sponsored Ads touches along the customer journey. This is particularly useful for full-funnel measurement – understanding how an upper-funnel DSP impression (like an OTT video ad) influenced later lower-funnel actions (like searches or product purchases).
For cross-channel marketers, cross-platform attribution remains a challenge. Amazon’s ecosystem is largely walled garden, meaning Amazon Ads attribution focuses on Amazon media’s impact on Amazon conversions (or on your site, if using DSP). If you want to see how Amazon DSP and, say, Google Ads work together, you will need to export data to your own analytics or use third-party attribution tools. Amazon’s own Amazon Attribution (not to be confused with attribution in DSP) is a tool that helps track how non-Amazon ads (like Facebook or Google campaigns) drive Amazon.com sales. It uses a similar 14-day last-touch model. While useful, it doesn’t merge with DSP data directly – they are somewhat siloed. To get a unified view of a customer who saw a Facebook ad and an Amazon DSP ad and then purchased on your site, an external analytics solution or data warehouse is needed. In absence of that, look at big-picture metrics: for example, if you increase spend on Amazon DSP, do you see overall search interest or direct traffic for your brand rise? Amazon’s own analysis noted that a well-executed DSP campaign can lead to stronger brand search volume and even nearly double total sales in some cases due to improved brand visibility.
Finally, Amazon has been rolling out predictive analytics and AI enhancements to measurement. One example (launched in 2024) is a predictive conversion modeling that attempts to connect ad impressions to downstream sales with more algorithmic attribution, even when direct tracking is sparse. These kind of AI-driven insights are aimed at uncovering latent impact (for instance, an impression that led to a purchase weeks later) while maintaining privacy. Advertisers should stay tuned to Amazon Ads updates, as features like these can further bridge attribution gaps and improve optimization.
To wrap up, here are some best practices and tips for making the most of Amazon DSP’s measurement capabilities:
Measuring what matters is paramount in Amazon DSP. With large budgets and multiple touchpoints at play, you need a clear view of how your ads contribute to outcomes. Amazon DSP offers powerful attribution tools – from 14-day last-touch reporting to advanced modeled and multi-touch analytics – to help advertisers connect the dots from impression to conversion. By understanding these tools and data, advertisers can continually refine their strategies: doubling down on effective tactics, fixing or dropping the underperformers, and ultimately driving better ROI.
In the fast-evolving world of Amazon Ads, staying on top of attribution trends (like the move toward multi-touch models) will keep you ahead of the curve. Always remember the context behind the numbers – use both Amazon’s reports and your own business metrics to judge success. When measurement is done right, Amazon DSP can be a game-changer, unlocking insights about your customers and significant growth for your brand. By aligning Amazon DSP’s strengths with a sound measurement strategy, you’ll ensure every programmatic dollar is well spent and accounted for.
