Amazon DSP expands conversion modeling to account for off-Amazon campaign results

Amazon this month announced an update to its Amazon DSP (Demand Side Platform), introducing modeled attribution for off-Amazon conversions for U.S. advertisers as of August 16, 2024.

This update aims to provide a more comprehensive understanding of campaign performance, particularly when direct measurement of conversions is not possible. Amazon DSP's modeled attribution helps advertisers track and evaluate the complete impact of their advertising investments across various supply sources, including anonymous inventory.

What Launched and Why It Matters

Amazon DSP now incorporates modeled attribution to account for off-Amazon conversions that are not directly measurable. This new feature integrates these modeled conversions with directly measured ones, presenting them as a combined result within campaign reports. The introduction of modeled conversions is essential for advertisers who need to assess the overall effectiveness of their advertising spend, especially in cases where traditional tracking methods may fall short.

The feature is particularly significant because it addresses gaps in data that can occur due to privacy restrictions or when conversions happen on anonymous inventory. Anonymous inventory refers to advertising spaces where direct user identification or tracking may not be feasible, thus making direct attribution impossible. By leveraging modeled conversions, Amazon DSP ensures advertisers do not miss out on evaluating the full value of their campaigns, which can include conversions that occur beyond the measurable scope.

Amazon DSP uses advanced modeling techniques to estimate these conversions, considering patterns and behaviors that are indicative of ad-driven actions. This approach helps advertisers maintain an accurate and holistic view of their campaign performance, allowing them to optimize their spending according to their key performance indicators (KPIs). Modeled attribution will become increasingly relevant as Amazon DSP continues to expand its ad delivery options, including the use of Ad Relevance tactics on anonymous inventory, where direct measurement may not always be achievable.

Where and How Advertisers Can Access Modeled Conversions

The modeled attribution feature for off-Amazon conversions is currently available to U.S. advertisers using Amazon DSP. Both managed-service and self-service advertisers are eligible to access this feature. Managed-service refers to advertisers who receive direct support and management from Amazon's advertising team, while self-service advertisers manage their own campaigns through Amazon DSP’s platform.

Advertisers can access these modeled conversions directly within the Amazon DSP campaign manager, the reporting center, and through the Amazon Ads API. The integration ensures that modeled conversions are reported alongside directly measured conversions, providing a unified view of campaign results. For advertisers who rely heavily on API data for reporting and analysis, the update automatically incorporates modeled conversions into all relevant metrics.

Amazon DSP’s API-specific endpoints will now include these modeled conversions, making it easier for advertisers to assess the total impact of their campaigns, even when conversions cannot be directly linked to specific ad interactions. The reporting process remains seamless, with no additional steps required from advertisers to access these combined conversion metrics.

Technical Insights into Modeled Attribution

Modeled attribution is part of Amazon DSP’s broader strategy to address the evolving digital advertising landscape, where data privacy and anonymous inventory play increasingly significant roles. Modeled conversions are not arbitrary estimates; they are calculated using sophisticated statistical models that infer the likelihood of conversions based on observable data points and historical patterns.

These models take into account various factors, such as the type of inventory, user interactions with ads, and broader market behaviors, to estimate conversions that could not be directly measured. For example, if an ad interaction occurs on an anonymous site or in an environment where user tracking is limited, Amazon DSP uses modeled attribution to estimate the potential impact of that interaction on conversion events. This technique ensures that advertisers can still make informed decisions about their campaigns without relying solely on direct tracking data.

The integration of modeled conversions into Amazon DSP reporting does not alter the direct measurement of conversions. Instead, it complements the existing data, offering a fuller picture of campaign performance. Advertisers can distinguish between directly measured and modeled conversions within their reports, enabling a detailed analysis of how their ads perform across different supply types.

Implications for Campaign Management and Strategy

The inclusion of modeled attribution has several implications for campaign management on Amazon DSP. By accounting for off-Amazon conversions that are not directly measurable, advertisers can optimize their ad strategies to ensure they are reaching their performance goals, even in less transparent environments.

The ability to evaluate modeled conversions alongside directly measured data helps advertisers maintain a consistent approach to performance tracking, even when dealing with anonymous or restricted inventory. As Amazon DSP continues to expand the use of anonymous inventory through Ad Relevance tactics, modeled conversions will provide critical insights into how these new ad placements contribute to overall campaign success.

Advertisers can leverage these insights to adjust their bidding strategies, targeting options, and overall budget allocations to maximize the effectiveness of their advertising efforts. For instance, campaigns that rely heavily on anonymous supply may benefit from focusing on modeled conversion metrics to ensure that performance evaluations remain comprehensive and accurate.

Key Features of the Modeled Attribution Update

• Availability: Modeled attribution for off-Amazon conversions is currently available to U.S. advertisers using Amazon DSP, including both managed-service and self-service users.
• Access Points: Modeled conversions can be accessed through the Amazon DSP campaign manager, reporting center, and the Amazon Ads API.
• Integration: Modeled conversions are reported together with directly measured conversions, providing a combined metric that represents the total impact of campaigns.
• API Inclusion: All API endpoints related to off-Amazon conversion reporting will now include modeled conversions, simplifying the reporting process for API users.
• Technical Approach: Modeled attribution uses statistical modeling to estimate conversions that occur on anonymous inventory or where direct tracking is unavailable.
• Strategic Impact: The feature enables advertisers to maintain a holistic view of campaign performance, especially as Amazon DSP expands ad delivery options in anonymous environments.

Key Facts

• Amazon DSP launched modeled attribution for off-Amazon conversions for U.S. advertisers on August 16, 2024.
• Modeled conversions are reported together with directly measured conversions, enhancing the overall view of campaign performance.
• The feature is available to both managed-service and self-service advertisers using Amazon DSP.
• Modeled conversions can be accessed through the Amazon DSP campaign manager, reporting center, and the Amazon Ads API.
• Modeled attribution helps advertisers measure the impact of campaigns across anonymous and directly measurable supply sources.
• Amazon DSP uses statistical modeling techniques to estimate conversions that are not directly measurable due to anonymous inventory or privacy restrictions.
• The integration of modeled conversions supports advertisers in managing KPIs and optimizing their ad strategies in less transparent environments.