Your achievements

Level 1

0% to

Level 2

Tip /
Sign in

Sign in to Community

to gain points, level up, and earn exciting badges like the new
Bedrock Mission!

Learn more

View all

Sign in to view all badges

India Hackathon: Predicting Anonymous Viewer Demographics with Ad Currency Optimization

Avatar

Avatar
Give Back 25
Community Manager
NimashaJain
Community Manager

Likes

43 likes

Total Posts

193 posts

Correct reply

1 solution
Top badges earned
Give Back 25
Seeker
Engage 1
Applaud 50
Validate 10
View profile

Avatar
Give Back 25
Community Manager
NimashaJain
Community Manager

Likes

43 likes

Total Posts

193 posts

Correct reply

1 solution
Top badges earned
Give Back 25
Seeker
Engage 1
Applaud 50
Validate 10
View profile
NimashaJain
Community Manager

24-10-2021

Authors: Atul Shrivastava, Prashant Dahiya, Praveen Kumar Goyal, and Jenny Medeiros

Banner Image.png

This post provides an overview of how we integrated Ad Currency Optimization within Adobe Experience Platform to streamline demographic predictions for anonymous users. This is the first of five exciting applications that resulted from our Adobe Experience Platform India Hackathon.

Advertising is being transformed by data and technology, but the lifeblood of successful video targeting remains the same: user demographics.

For publishers of TV streaming content, most of their users are anonymous. And without demographic data on their viewers, they can’t deliver relevant advertising to them. These missed targeting opportunities not only submit their viewers to a barrage of irrelevant ads but can also cost the publisher between 40–45% of their ad revenue.

Adobe Primetime is a cloud-based platform that helps digital video publishers increase their advertising and subscription revenue. One of its key components is Ad Currency Optimization, a patented demographic model that uses machine learning to predict age and gender-based on viewing behavior, so publishers can deliver better-targeted ads.

During the India Hackathon, members of the Adobe Primetime team collaborated with the local Adobe Experience Platform Engineering team to repackage Ad Currency Optimization as an intelligent service within Adobe Experience Platform, resulting in:

  • Streamlined data processing
  • Seamless activation for customers
  • Near real-time audience segment delivery
  • Shorter time-to-value for customers

In this post, we describe what makes Ad Currency Optimization so powerful for publishers, how we integrated it into Adobe Experience Platform, and what we plan on doing with it next.

Predictive intelligence for effective advertising

Media and entertainment (M&E) publishers, such as broadcast networks, want to target their ads better and achieve greater ROI on their digital television content.

Much like with traditional television, digital publishers sell their ad slots against demographic guarantees (e.g. females between 18–49 years old), and only pay for ad impressions that land on that specific demographic. With no data to target anonymous users, however, most of the ad campaign impressions fall on users outside of the desired demographic, resulting in a lose-lose situation for the publisher, advertiser, and also the viewer.

This is where Ad Currency Optimization comes in. Ad Currency Optimization uses machine learning to analyze how the viewer interacts with content to deliver predictive audience segments, which are optimized against third-party demographic measurement providers. Lastly, the segment is delivered to the customer so they can finally target their “mystery viewers” with personalized ads.

With Adobe Primetime, Ad Currency Optimization has successfully led to a 32–48% increase in revenue potential for big-name customers, including major US TV networks. Now we aim to extend the benefits of predictive audience segments for Adobe Experience Platform customers, too.

Bringing Ad Currency Optimizer to Adobe Experience Platform

For context, the data flow for Ad Currency Optimization is as follows:

Figure 1: Diagram of the existing data flow for Ad Currency Optimization.Figure 1: Diagram of the existing data flow for Ad Currency Optimization.

  • Adobe collects viewing data from the publisher’s Adobe Analytics account, including what shows viewers have been watching, what time of day they watch them, and the devices they are using.
  • Adobe Sensei, our artificial intelligence and machine learning technology, analyzes the data against a third-party measurement provider to learn which behaviors predict viewers’ demographic characteristics.
  • Adobe creates predictive audience segments that are optimized for specific demographics provided by the customer (e.g. females between the ages of 18 and 49).
  • The audience segments are placed in the customer’s Adobe Audience Manager account, where they can export the segment into their ad server for effective audience targeting.

For the hackathon, the scope of our challenge was to integrate Ad Currency Optimization within the Adobe Experience Platform Pipeline, resulting in the following workflow:

Figure 2: Diagram describing the data flow for Ad Currency Optimization as a service in Adobe Experience Platform.Figure 2: Diagram describing the data flow for Ad Currency Optimization as a service in Adobe Experience Platform.

It begins by sourcing viewership data from the Adobe Analytics data in XDM format already streaming into Adobe Experience Platform. This data is then fed into Ad Currency Optimization — deployed on Adobe Experience Platform Data Science Workspace — where the demographic model predicts the viewer’s age and gender based on their viewed content and behavior.

These predictions (i.e. probability scores) are then added to the dataset within Adobe Experience Platform. The scores are also exported and sent into a feedback loop with a third-party measurement provider that validates the accuracy of our predictions.

Once the probability scores have been refined, the dataset travels through the Unified Profile Service within Adobe Experience Platform, where they are grouped into audience segments based on their common traits. Lastly, we use real-time destinations in Real-Time Customer Profile to push these segments to the publisher’s ad server for our customers to access and act upon.

Benefits of Ad Currency Optimization in Adobe Experience Platform

When comparing the effectiveness of Ad Currency Optimization with its standalone pipeline versus Ad Currency Optimization within Adobe Experience Platform, we can proudly highlight the following benefits:

Reduced operational cost

With the power of machine learning in Ad Currency Optimization, we can cut the cost and time-consuming efforts spent on backend services, such as extracting and cleaning data. By automating these services, the team can focus instead on improving our core offerings: a demographic prediction model and validation with measurement vendors.

Near real-time segment transfer

With Ad Currency Optimization, segment transfer happens in near real-time so publishers can instantly monetize the segment. Previously, the process of ingesting the segment and pushing it to the customer’s ad server lasted an entire four days, which was essentially lost time and ad revenue for the publisher.

Shorter time-to-value

Integrating Ad Currency Optimization within Adobe Experience Platform reduces the steps needed to onboard a customer to the service.

Instead of spending 4–5 weeks understanding, exporting and normalizing a customer’s Adobe Analytics data before feeding it into the predictive model — or setting up an Adobe Audience Manager destination to activate the segments in their ad server, we can successfully onboard a customer within just one week so they can better target their viewers sooner.

Implementation challenges

Implementing Ad Currency Optimization within Adobe Experience Platform was not without its speed bumps.

Our main challenge was the limited availability of the real-time segment destination since it could only be accessed by customers with Real-Time Customer Profile enabled. Furthermore, it skipped over popular ad servers in the M&E space (e.g. Freewheel).

As a solution, we exported the segments from Adobe Experience Platform to Adobe Audience Manager first. Next, we pushed those segments into the customer’s ad server using Adobe Audience Manager’s segment destination (batch transfer) capabilities. This workaround ensures segments are accessible for Adobe Experience Platform customers.

What’s next for Ad Currency Optimization?

Merging Ad Currency Optimization and Adobe Experience Platform creates new ways for publishers to harness data and machine learning for better ad targeting — beyond age and gender demographics.

This technology also opens many new audience targeting opportunities. For example, it can be applied to predict and target the most likely audiences that would take action in response to an ad, such as downloading a mobile app or visiting a brand’s website.

Given the compelling benefits of Ad Currency Optimization and our continuous progress with this innovative service, it has become a top priority for the Adobe Primetime product team and is expected to fully roll out in 2020.

To learn about what other exciting applications we have in store for Adobe Experience Platform, read our India Hackathon summary.

Follow the Adobe Experience Platform Community Blog for more developer stories and resources, and check out Adobe Developers on Twitter for the latest news and developer products. Sign up here for future Adobe Experience Platform Meetups.

Resources

  1. Adobe Experience Platform — https://www.adobe.com/experience-platform.html
  2. Adobe Primetime — https://www.adobe.com/marketing/primetime.html
  3. Ad Currency Optimization (in Adobe Primetime) — https://www.adobe.com/marketing/primetime/currency-optimization.html
  4. Data Science Workspace — https://www.adobe.com/ca/experience-platform/data-science-workspace.html
  5. Adobe Sensei — https://www.adobe.com/sensei.html
  6. Adobe Audience Manager — https://www.adobe.com/analytics/audience-manager.html
  7. Adobe Analytics — https://www.adobe.com/analytics/adobe-analytics.html

Originally published: Apr 16, 2020