supertype logo dark (wide)
supertype logo dark (wide)
zoomd

Predictive Analytics in programmatic media buying

Programmatic audience creation through a bespoke machine-learning-as-an-API service, so media buying on RTB (real-time buying) exchanges are timely and with a stream of pre-qualified audience.

At a Glance

Challenges

  • Sifting through terabytes of data across different servers for segmentation tasks is practically challenging
  • Unable to pre-emptively omit devices that are unlikely to respond to an offer, resulting in wasteful bids on low- or non-converting impressions on the ad exchange

Benefits

  • Device-level targeting driven by machine learning algorithms on a bespoke API service result in 60% less costs than would otherwise be necessary
  • A 23% increase in response rate with an adaptive and predictive segmentation algorithms hosted on Supertype’s servers

Knowing who to reach, and exactly when

Challenges

Zoomd is connected with the world’s largest ad exchanges and operate an industry-leading demand-side platform (DSP). In an effort to deliver media buyers an outsized ROI, it needs to incorporate precise device-level targeting based on advanced segmentation and machine learning techniques — while omitting devices that are unlikely to respond to a certain campaign offer.

Solutions

  • Precision in media buying on its RTB platform

    Zoomd offers media buyers of its DSP platform a high curated set of audience(s), improve relevance and media buy effectiveness through layering its real-time bidding capabilities with high-precision audience scoring and segmentation

  • Implementing a 'negative' list to improve CTR

    Zoomd connects to the API server developed by Supertype's engineers. Its DSP thus pre-emptively omit users predicted to be unresponsive to an offer, with media budget allocated elsewhere, improving overall CTR by 23% over the control group

  • Improved Lifetime value (LTV)

    By proactively identifying impressions that were likely low- or non-performing, Zoomd is able to allocate their bids more competitively toward impressions that are scored favorably, thus netting a higher lifetime value and ROAS (return on ad spend) on average

Approaches

Programmatic media buying powered by a Predictive and Adaptive Audience Creation Egine

Seamless, multi-faceted audience scoring

Supertype developed an API service that is connected to Zoomd’s data infrastructure, which intelligently score each device based on their likelihood to respond, or not respond, to a campaign offer. 

The algorithm takes into account factors such as the device owner’s pre-existing affinity toward the different media types, genres, attention spans to generate anonymous clusters of users, from which the bidding team would select or deselect from their audience set-up.

Cost-effective Machine Learning as a Service

Supertype leverages on Amazon S3 (Simple Storage Service) and Amazon Lambda in the development of this solution. This eliminates the need for Zoomd to maintain or manage any server infrastructure, instead relying on deployed code (the machine learning algorithm) being triggered as new events are streamed into storage. 

Implementing serverless technology to perform prediction as and when needed reduces client’s cost by 60% compared to conventional methods. 

Adaptive Non-target List and Fatigue Scoring

Zoomd’s technology team and media buying team works with Supertype’s machine learning engineers to incorporate models that would pre-emptively and selectively omit users unlikely to respond to ad (“ad fatigue”) from its media buyers, leading to a 23% increase in responses rate (e.g. Click-through rate) compared to industry standards.

LTV (Lifetime value) Prediction

Pre- and post-bid analysis offers key opportunities to study the key factors that correlate to an uptick in response rate, occurrence of downstream events (“in-app” or “post-installation”) via 3rd party attribution SDK, and other engagement metrics. These are then treated as coefficients in the training of LTV (lifetime value) predictive models.

Value Add with Supertype Summary

Levelling up Zoomd's Client Success team

Working with clients for their programmatic media buying business require Zoomd to be prompt and proactive in their communication and client service. Supertype's PDF report generation as a service take this up a notch by allowing Zoomd' client success team to create Zoomd-branded, visually stunning, AI-powered reports and analysis in matters of minutes, not days.

Supertype Summary
supertype summary
supertype summary
supertype summary
supertype summary pdf generation
Previous
Next
The programmatic report generator [Summary] that Supertype created was very helpful in helping us engage with our mobile app clients, and in opening conversations. Most of all, the charts it generate looks gorgeous and the topic identification model is spot on!
Yair Yaskerovitch
COO, Zoomd Technologies (ZOMD.V, ZMDTF)

Contact Us

Zoomd

Want more case study?

This case study, along with others in our archive is also available as a PDF. To schedule a call with us and learn more about our past projects, leave your contact method below.

Your work email, so we can reach you. We respect your privacy and do not engage in any spamming activities.

Inquiries

s@supertype.ai

Back to homepage

More content from Supertype