Published February 10, 2022

Learn Why MediaMath Selected Ocient for Next Generation Campaign Forecasting

MediaMath unlocked $100 million in revenue with a custom Ocient analytics solution

At a Glance:

  • 9-12 months of internal engineering team time saved
  • $100 million opportunity for new revenue growth unlocked
  • 50% savings in time and cost to get to market

Top Highlights:

  • Delivered rapid analysis of 2-5x more data on a multi-petabyte dataset, generating accurate, granular insights for advertising clients to better forecast future campaigns
  • Reduced total cost of ownership by 50% by decreasing overall system footprint and streamlining resource usage via Ocient’s end-to-end hyperscale data analytics solution
  • Optimized return on ad spend for customers enabling MediaMath to retain key accounts and be first to market to capture new customers and revenue growth

After more than a decade as a leading advertising technology pioneer, New York City-based MediaMath needed to innovate quickly in the campaign forecasting space. With more than 3,500 advertisers relying on their demand-side platform and modernized ecosystem to solve complex marketing problems, MediaMath was already helping their clients deepen customer relationships across screens and around the world. But their advertising clients wanted the ability to scientifically leverage more granular campaign insights to forecast more accurately for future campaigns and reduce wasted or poorly targeted impressions.

The MediaMath demand-side platform processes six million bid opportunities per second, generating up to 12 petabytes of new records each day. As critical columns are analyzed, there can be over 7 trillion data points that may need to be evaluated in a forecast query. In the past, the industry standard was to deliver estimates of potential campaign reach, impressions, and spend by using smaller samples of data derived from data rollups. While customers could easily see approximate results, general trends, and big-picture analysis, they were missing those needle-in-a-haystack insights into their advertising campaigns that are so essential for maintaining a competitive edge.

MediaMath was uniquely positioned to be the first to market with a solution that would allow advertising clients to plan their campaigns with higher resolution insights, which in turn would help them make more accurate predictions across engagement points, including connected devices like smart TVs – all with less waste and higher ROI. This solution would not only ensure they retained key accounts, but also unlock a huge potential for new business. It would operate at a huge scale and demand a huge increase in storage footprint and cost. It could also take a team of their best engineers up to a year to build.

Replacing costly Spark and Hadoop jobs

Chris Ingrassia, principal engineer, and the technical leadership at MediaMath started looking for the right tools to create a solution that could analyze billions of records in seconds. Their existing implementation of Spark and Hadoop simply couldn’t scale to handle the amount of data they were looking to extract, transform, and load. Scaling these legacy solutions would not only be prohibitively expensive but would also cause data pipeline problems, all without providing the ability to run the full slate of complex queries required for high-resolution campaign insights. And with the explosion of data volumes across platforms, MediaMath knew they would need a solution that could handle their data needs now and into the future.

That’s when MediaMath connected with Ocient.

“When I first heard about Ocient’s data warehouse platform, the technology was instantly impressive,” said Chris Ingrassia.

Ocient’s technology delivered low latency analytics and continuous loading and transformation of raw, semi-structured bidder log data. And with Zero Copy ReliabilityTM leveraging parity coding for enhanced data availability, MediaMath could reduce their storage footprint by more than 50 percent.

Ocient had the right technology, but MediaMath needed more than just a data warehouse and analytics platform. They needed an end-to-end production grade campaign forecasting solution. Ocient offered a partnership where MediaMath could not only implement this time- and cost-saving technology, but also rely on the Ocient team to build and test the entire custom solution. It was exactly what they needed to deliver their cutting-edge solution to the market before any of their competitors.

Building faster and smarter with Ocient’s hyperscale data solutions experts

The Ocient and MediaMath teams worked closely together to design, test, and implement the end-to-end campaign forecasting solution. Ocient replaced MediaMath’s legacy combination of disparate solutions, including Spark and Hadoop processing jobs and ETL steps with a unified platform that streams and transforms data at ingest.

 

MediaMath replaced a costly collection of Spark and Hadoop jobs with an efficient Ocient platform that streams and transforms data at ingest.

Ocient’s solution delivered enhanced efficiency by creating three service classes for users within the analytics platform. Managing workloads in this manner allowed for seamless resource sharing between mission-critical tasks, medium priority background jobs, and ad-hoc queries. Three workload service classes ensure that storage and data usage are optimized based on exactly how MediaMath needs to use each piece of data.

In addition to streamlining the overall analytics flow, this new setup also enables rapid development on top of Ocient, so MediaMath can add new capabilities without a heavy data engineering investment. This opens the door for new and innovative features, drastically reducing the time to market for feature innovations.

Delivering seamless implementation and hosting in OcientCloud™

In a matter of months, the Ocient solutions architects and database engineering team delivered a production-ready solution with minimal operational effort from MediaMath.

“We were able to build a petabyte-scale application with only 20% of the internal engineering resources it would typically take,” said Ingrassia. “Without Ocient, I would have needed tens of engineers to build this solution over 9-12 months. Ocient helped us get to market faster and simplified the process along the way.”

Because the Ocient solution doesn’t require complex tuning or index management, MediaMath won’t need to invest in additional headcount to maintain their solution. It just works.

Best of all, the entire solution is managed and deployed in the OcientCloud™.

“When I compared Ocient’s hosted cloud solution to the largest cloud providers, Ocient was far more cost effective,” said Ingrassia. “Plus, their fully hosted service already met our data protection and privacy requirements, including GDPR, CCPA, and SOC2.”

Operating in OcientCloud was a significant factor in lowering MediaMath’s total cost of ownership for their new forecasting solution. With a revolutionary Compute Adjacent Storage Architecture™ (CASA) designed to maximize compute at the IO layer on NVMe SSDs, and Zero Copy Reliability™, which delivers reliability and availability without replication, Ocient reduced the storage footprint by more than 50 percent. It’s a huge savings for MediaMath and provides much-needed opportunities to take on more data as their business and advertising data grow.

Now, MediaMath is delivering deeper forecasting insights for their clients, which is already fueling new growth and opportunities for marketers and advertisers to reach their audiences in more relevant and contextual ways.

“Our customers rely on MediaMath to innovate continuously and bring more competitive insights across more screens and platforms faster than ever,” said Anudit Vikram, chief product officer at MediaMath. “Working with Ocient enabled us to get to market months faster. We can now grow our market share and pursue massive new business opportunities while reducing costs and eliminating the resource-intensive process of customizing our own solution internally.”

What’s next for MediaMath and Ocient? With 24/7 support and comprehensive system management as part of their OcientCloud implementation, the team will have Ocient’s full collaboration to take on whatever new challenges come their way.

“There’s nothing the team at Ocient can’t or won’t tackle,” said Ingrassia. “They were easy to work with and architected a system that met our requirements at the time while also future proofing the system to expand and take on more as our needs grow.”

Whether you’re in AdTech or another industry, Ocient prides itself on partnering closely with customers to tackle their toughest hyperscale data challenges. Contact sales@ocient.com for more information or contact us via the website.