AWS for Industries

Integrating CDP with Amazon Marketing Cloud to drive better Ad campaigns

There is customer data across more sources today than ever—from email, websites, and mobile apps, to physical stores, social media platforms, and more. Such data, which a company directly collects from its customers, is referred to as first-party data or signal (1P signal). Creating actionable insights, and next-best actions for specific users or segments, from unified data has become a paramount concern. Many companies are using off-the-shelf Customer Data Platforms (CDP) or in-house built CDP solutions on Amazon Web Services (AWS) to ingest and unify 1P signals, and to create the actionable insights for activating Ad campaigns across channels.

Today, a growing number of companies are offering enterprise CDPs on AWS. Our previous blogs A CDP for CPG?! and Why Retailers Need a CDP to Power Their Customer 360 Strategy outline reasons for why a CDP is important to retailers and consumer packaged goods companies.

In this blog, we will focus on a leading CDP provider Treasure Data, which built a successful integration with Amazon Marketing Cloud (AMC), the clean room solution of Amazon Ads, to enable new ways of turning 1P signals into insights. By utilizing the secure, privacy-safe, and cloud-based clean room infrastructure AMC offers, advertisers are able to quickly perform analytics using pseudonymized signals from both Amazon Ads and the advertiser’s own inputs from CDP, and generate actionable insights to inform advertising strategy. We will also look at initial results and discuss how native integrations between CDPs and Amazon Ads can help advertisers and marketers better understand their audience through a single pane of glass without leaving the CDP user interface.

The following diagram describes how companies can unify signals, generate insights, and activate custom audiences created based on those insights through the integration.

Image 1 - 1P + AMC Signal Unification, Insight Generation, Media Activation workflow

Image 1: 1P + AMC Signal Unification, Insight Generation, Media Activation workflow

The integration between Treasure Data CDP and AMC allows advertisers to use Amazon Ads to pass strategic CDP segments from Treasure Data in a pseudonymized form to AMC. Advertisers can then combine the CDP inputs with Amazon Ads signals and obtain unique insights on topics such as media impact, audience segmentation, segment overlap, and customer journeys in a privacy-safe manner.

These insights from AMC can be leveraged by Treasure Data customers to learn more about their audiences and further refine their media strategy. For example, by overlapping 1P signals with Amazon Ads reach, an advertiser is able to determine whether they reached incremental potential customers. Amazon Ads advertisers who are also Treasure Data customers can activate against enhanced audiences in Amazon demand-side platform (Amazon DSP) to maximize the performance of their campaign objectives. Amazon DSP is a demand-side platform that allows you to programmatically buy ads to reach new and existing audiences on and off Amazon.

Note: Amazon Marketing Cloud only accepts hashed/pseudonymized information. All information in an advertiser’s AMC instance is handled in strict accordance with Amazon’s privacy policies, and an advertiser’s own signals cannot be exported or accessed by Amazon. Advertisers can only access aggregated, anonymous outputs from AMC.

Image 2 - Solution architecture detail and signals flow

Image 2: Solution architecture detail and signals flow

Solution Details

The following events and signal flow are mapped to the image above:

  1. Advertiser activates a campaign using Treasure Data’s CDP integration with Amazon DSP.
  2. Ad-attributed engagement signals received by AMC (views, clicks, purchases, and more).
  3. Treasure Data’s schema-less architecture ingests and integrates 1P customer data across multiple sources to build unified customer profiles. Data is streamed or batched using 250+ pre-built connectors at scale (a typical CDP functionality).
  4. Advertiser’s Treasure Data administrator installs the AMC integration by working with the Treasure Data account team. They perform a one time on-boarding using Login with Amazon, the same credential used to login to Amazon Ads solutions (Amazon DSP, AMC). Treasure Data saves the OAuth token for future use.
  5. Advertiser creates multiple audiences using Treasure Data Audience Studio user interface. Audience segments are loaded in SaaS tenant’s Amazon Simple Storage Service (Amazon S3) buckets. It activates the Treasure Data workflow engine.
  6. Treasure Data Workflow Engine, using AWS Fargate, prepares 1P audience inputs. Note, personally identifiable information (PII) fields are hashed using SHA-256.
  7. Pseudonymized audiences are sent from Treasure Data CDP to AMC by using AMC API.
  8. AMC triggers query workflow and insight generation. Example insights include top performing 1P audiences by media channel, top Amazon segments by brand loyalty program status, and more. More than 20 out-of-the-box query workflows can be run as part of the Treasure Data AMC application.
  9. (9.a) Advertiser insight outputs are in aggregated and anonymous format received by Treasure Data. (9.b) For example, an advertiser can have insights into who already bought the product from the brand, and can therefore build a new custom audience directly in AMC. They can then activate a cross-sell campaign to reach these audiences in Amazon DSP (future capability).
  10. Insights received from AMC are loaded in Treasure Data visualization user interface (UI) and available for further analysis with 1P insights.
  11. Advertiser further analyzes audience insights from AMC and understands specific Amazon segments that convert better.
  12. With 1P and AMC combined insights in CDP, advertiser is able to create, for example, re-marketing campaigns and execute a/b tests to optimize their Return on Ad Spend (ROAS).

Amazon DSP usage is a prerequisite for an advertiser to use AMC and Treasure Data’s AMC application. Additionally, to activate on AMC audiences through this solution, the advertiser requires their ADSP advertiser Amazon customer-facing ID (CFID) mapped to the AMC instance from their respective ADSP entity.

Once the advertiser creates an audience using Treasure Data Audience Studio, reports can be generated to get insights from the combined signals from AMC. For example, the advertiser can upload email campaign signals and compare this against AMC campaign KPIs. A few other common use cases include:

  • Audience overlap: Understand the overlap between CDP audiences and those engaged with their Amazon Ads campaigns.
  • Audience attributes: Assess the Amazon audience segments their CDP audiences belong to.
  • Off-Amazon ad impact: Measure the impact of Amazon Ads campaigns on direct-to-customer engagements and conversions.
  • Customer journey: Evaluate customer activities across Amazon Ads channels as well as brand-owned channels such as email and websites.

Advertisers can take actions based on these insights and plan for future campaigns in Amazon DSP. The outcomes provided from these analyses are aggregated and not available at the user level.

Example Use Case: Insights Based on Loyalty Status

The input segment provided by the advertiser could contain advertiser-related information that is not already available on AMC. For instance, the advertiser may like to analyze their brand loyalty program information against the Amazon Ads audience segments. This analysis may generate insights into what percentage of known loyal customers are buying from Amazon stores and what Amazon Ads audience segments they belong to. The provided inputs can be a dimension or fact dataset. Below is an example of the anonymized inputs from Treasure Data that were analyzed and joined with Amazon Ads signals in AMC.

Input ready for AMC with hashed PII:

Image 3 - Example 1P segment with loyalty_status

Image 3: Example 1P segment with loyalty_status

Note: All Personally Identifiable Information (PII) fields in Image 3 is hashed using SHA-256 before uploading to AMC.

Image 4 - Example of Loyalty Status overlap with Lifestyle (LS) Foodies insight

Image 4: Example of Loyalty Status overlap with Lifestyle (LS) Foodies insight

For all overlapped users, analysis can be done to find out the lifestyle segments audiences belong to and patterns on loyalty member status.

Furthermore, the advertiser can perform this for individual segments and market to these segments in future campaigns on AMC. Advertisers can understand total purchases, money spent, add-to-cart, and detail page views for the segment at an aggregated level.

The advertiser can also see which segments of their major customer base have not been reached by their Amazon DSP campaigns, and take action to reach those segments in the next campaign.

Image 5 - Example analysis for future campaigns

Image 5: Example analysis for future campaigns

The example in Image 6, illustrates an analysis for insights such as top purchase segments, ad impressions by gender, or reach by states—visualized in the CDP user interface.

Image 6 - Treasure Data Insight Analysis

Image 6: Treasure Data Insight Analysis

With over 20 out-of-box templatized query workflows being available in AMC, rich and in-depth analysis is possible for the advertiser and agency to generate always-on insights and drive ROAS optimization.


“This turnkey integration with Treasure Data and Amazon Marketing Cloud, that’s supported by AWS, helps mutual customers to deliver more relevant ad experiences and gain customer insights using rich event-level signals”, said John Grundy, Head of Product Management at Treasure Data. “Our marketing and advertising customers are constantly looking for ways to unify signals, simplify workflows, and optimize their advertising spend—this joint solution allows them to exactly do that.”

If you are an advertiser already using Treasure Data CDP and AMC, you can get started quickly by activating the AMC integration in the Treasure Data UI.

If you are a software vendor, system integrator, or brand advertiser considering building this integration, contact your AWS team (typically a PDM or Account Executive).

Customers storing their insights on AWS are looking for streamlined ways to analyze and draw insights with Amazon Ads signals through AMC—we have two joint solutions that help:

  1. AMC Uploader from AWS helps Amazon Ads customers seamlessly upload customer signals into their AMC instance, without dedicating IT resources to build and support the upload workflows.
  2. AMC Insights on AWS helps advertisers and agencies running campaigns on Amazon Ads to quickly deploy AWS services to analyze and visualize reporting from the AMC API.

Together, these two solutions help customers reduce deployment time and streamline ingestion into AMC. In either case, advertisers will need to have an AMC instance API key and an AWS Account available to get started.

AWS Partner spotlight

Treasure Data helps enterprises use all of their customer data to improve campaign performance, achieve operational efficiency, and drive business value with connected customer experiences. The Treasure Data Customer Data Cloud, suite of customer data platform solutions, integrates customer data, connects identities in unified customer profiles, applies privacy, and makes insights and predictions available for Marketing, Service, Sales and Operations to drive personalized engagement and improve customer acquisition, sales, and retention. If you are an existing Treasure Data customer, you can activate AMC integration in the Treasure Data UI.

Treasure Data CDP in AWS Marketplace >>

Ronak Shah

Ronak Shah

Ronak Shah is a Senior Partner Solution Architect with the AWS Industry Vertical team based in the New York City area. He works with ISVs in the Retail and CPG verticals to co-innovate on AWS. He is interested in finding new trends in retail and building innovative solutions in the areas of digital commerce, supply chain, customer experience and marketing technology. Outside of work, he volunteers at scouting and local debate competitions.

Ginni Malik

Ginni Malik

Ginni Malik is a Data & ML Engineer with AWS Professional Services. She assists customers by architecting enterprise-level data lake solutions to scale their data analytics in the cloud. In her spare time, she is a travel enthusiast.