Nielsen empowers agencies with Advanced Audience media planning and measurement on AWS Clean Rooms

With the changing landscape of loss in signal addressability and with companies prioritizing consumer privacy, data protection, and data control, Nielsen is using privacy-enhancing technology on Amazon Web Services (AWS) like AWS Clean Rooms—a service that helps you collaborate with your partners without sharing raw data—to help agency customers better understand the impact of their advertising. The Nielsen ONE platform validates big data with Nielsen’s people-based panels and device recognition to provide accurate granular insight into impressions, reach, and frequency across platforms. Advanced Audience measurement in Nielsen ONE lets Nielsen’s customers use their own first-party data on Nielsen’s platforms for deduplicated, cross-channel, unified audience insights beyond the traditional age and gender demographic segments.

Through Nielsen ONE’s use of AWS Clean Rooms to integrate first- and third-party data, brands and agencies can gain unique insights about audience reach and campaign effectiveness at levels of precision not possible without first-party data—all without having to move, copy, or share their first-party underlying data with Nielsen. This creates a better customer experience by letting customers easily and securely analyze and collaborate on their collective data with Nielsen. With Nielsen ONE Advanced Audiences, brands and agencies gain detailed insights into where they should invest their media dollars to better personalize the customer experience and drive increased business value.

AWS Clean Rooms is a service that helps companies and their partners to more easily and securely analyze and collaborate on their collective datasets, without sharing or copying each other’s underlying data. Using AWS Clean Rooms, you can create a secure data clean room in minutes and collaborate with any other company on AWS to generate unique insights to help inform campaign planning, targeting, measurement, and optimization.

Nielsen saw a need to automate onboarding for partners when collaborating using AWS Clean Rooms. The company wanted to maintain a standard process and data schema for future collaborations. At the same time, it realized the need to maintain flexibility for data scientists while authoring clean room queries.

This post outlines how Nielsen implemented governance controls to let data scientists safely iterate on their AWS Clean Rooms analysis templates. It also shows how Nielsen improved onboarding for its partners by automating the configuration of an AWS Clean Rooms collaboration with Nielsen. Using a universal template, an automated infrastructure-as-code (IaC) deployment pipeline generates a collaboration in AWS Clean Rooms through a single technical process flow. This lets Nielsen remove configuration responsibility from its clean room users without compromising flexibility and consumer privacy.

Overview of the solution

The challenges

Nielsen needed to address several challenges:

• Sensitive data: Developing new analyses required the ability to iterate on sensitive datasets. This required controls to safeguard the datasets.
• Multiple skill sets: Developing clean rooms queries required expertise in data engineering, managing AWS infrastructure, and data science. Coordinating across multiple teams slowed time to insight.
• Manual processes: Creating clean rooms, often requiring the same schemas, involved multiple steps. Doing them manually was time consuming and prone to errors, and it required manual creation for each collaboration.

The automated provisioning pipeline

AWS provides the ability to create AWS Clean Rooms collaborations using AWS CloudFormation-, which lets you model, provision, and manage AWS and third-party resources. AWS CloudFormation is a powerful IaC framework that customers can use to automate infrastructure deployments and manage their resources across their organizations. Nielsen used AWS CloudFormation for its automated provisioning pipeline.

The pipeline is designed to remove administrative burden for Nielsen’s agency customers during the onboarding process of an AWS Clean Rooms collaboration. It comes in after Nielsen and the customer complete a digital form during the data agreement process. This form captures all the necessary details required for the AWS Clean Rooms

configuration. To configure a collaboration environment in AWS Clean Rooms, the collected form data is fed into an AWS CloudFormation template to provision and manage infrastructure resources in a consistent and repeatable manner.

Figure 1. Provisioning steps for a new collaboration

Prerequisites for collaboration

Before running the automated provisioning pipeline, Nielsen verifies that certain prerequisites are met:

1. Existing tables: Both Nielsen and the agency must have existing tables in AWS Glue—a serverless data integration service—that point to their respective buckets in Amazon Simple Storage Service (Amazon S3)—an object storage service. These buckets contain the data to be collaborated on in the clean room.
2. Data agreement: A formal data agreement must be in place between Nielsen and the agency, outlining the terms and conditions for privacy-enhanced data collaboration, including which fields to include and agreed-upon schemas.

Figure 2. Solution components

1. The input form values are populated into a template that is used in Nielsen’s provisioning pipeline.
2. The engine runs the first template.
3. This creates the collaboration.
4. The agency receives a membership invitation, which it will accept on its side.
5. The engine creates configuration tables, then associates the tables to the collaboration.
6. The agency follows its own processes to create and associate its configuration Tables.
7. The agency can then run an analysis template, which creates advanced audience data that is surfaced in Nielsen ONE Advanced Audiences.

Data scientist empowerment

Nielsen recognizes the importance of empowering its data scientists to explore and experiment within a data clean room collaboration while authoring queries against the collective datasets that include their customer’s first-party data. To achieve this, Nielsen implemented an automated approval system based on tagging AWS resources. This system lets Nielsen’s data scientists play with templates in an one-time manner. Analysis templates in AWS Clean Rooms let customers define parameters to help them reuse queries. To maintain control and prevent accidental modifications or deletions, Nielsen enforces the following rules:

1. Data source and workflow: Data going into the clean room for analysis is sourced from Nielsen’s Media Data Lake and associated to AWS Clean Rooms. Data scientists do not need to use AWS Glue directly.
2. Analysis template changes: Data scientists can create and test templates against nonproduction data while developing queries, but they cannot use these templates with Nielsen’s extract, transform, load (ETL) scheduler until they apply an AWS tag to the template, which effectively approves (promotes) it for use in the next scheduled run.
3. Analysis template promotion: Nielsen and the agency review templates. Upon mutual agreement, a data scientist applies the required AWS tag to a template. With the application of the tag, the template is effectively flagged as promoted. This makes it eligible for a scheduled run against production data.
4. Analysis template scheduling safeguards: To verify that clean room queries cannot be regularly scheduled until promotion, roles that are associated with task orchestration within AWS Identity and Access Management (AWS IAM)—a service used to securely manage identities and access to AWS services and resources—are not capable of running analysis templates without a promotion marker.
5. Data validation: Before the template is run, data validation queries are run against the datasets to verify that collaborators loaded in the right data.

This approach strikes a balance between empowering data scientists to explore and experiment while verifying that promoted templates remain untouched and that scheduled joins are not disrupted.

Figure 3. Data scientists tag templates that are approved for use

1. During development, data scientists design queries and test their templates as needed.
2. When templates are ready to be used for a scheduled run, they are tagged to be “promoted.”
3. The ETL scheduler uses the promoted template on the next scheduled run.
4. Resulting data outputs to the designated Amazon S3 bucket.

Conclusion

Nielsen’s innovative automated provisioning pipeline demonstrates Nielsen’s commitment to simplifying data collaboration practices while enforcing data privacy. It also demonstrates how using IaC on AWS as well as custom automation can streamline data collaboration processes.

By abstracting the configuration mechanics and using automated processes, Nielsen streamlines the onboarding experience for its agency customers. Additionally, its tagging system for template management empowers data scientists to collaborate effectively while maintaining control over promoted analysis templates and scheduled data joins.

If you are interested in adopting AWS Clean Rooms technology contact our team of experts here, or get started using the service today.

About Nielsen

Nielsen shapes the world’s media and content as a global leader in audience measurement, data, and analytics. Through its understanding of people and their behaviors across all channels and platforms, Nielsen empowers its clients with independent and actionable intelligence so they can connect and engage with their audiences—now and in the future. Nielsen operates around the world in more than 55 countries.