Ad Campaigns, Optimized: Big Data Helps DSPs Translate Insights into Action

By Gidon Ben-Zvi


Not long ago, marketing teams would spend days analyzing mounds of Excel documents that contained campaign data, reading the tea leaves in an attempt to connect desired actions to the right customers.

Today, demand-side platforms (DSPs) are leveraging machine learning in a real-time environment to find data connections previously hidden and translate these insights into actionable campaign decisions.

Indeed, real-time bidding and DSPs have become major points of interest in Adtech, since more than 30% of total ads are going through demand-side platforms.

Growing Trend: The DSP market size in is expected to reach 24.5 billion yuan


While the Ad Exchange is the body, the DSP acts as the brain, which needs thousands of billions of neurons to provide marketers with the capacity to reason, comprehend, and make decisions. This is where data management platforms (DMPs) being developed by Big Data vendors enter the picture, since they give DSPs access to tons of information and data. A DMP is basically a platform that uses Big Data to first define a target audience and then provide pertinent information, which will ultimately be turned into actions that can be planned, delivered and measured in real time.

Along with the growth of Real-Time Bidding (RTB), demand-side platforms are a perfect tool to fuse the automation of the buying process with the efficiency of audience segmentation. Big Data platforms are helping optimize DSP efficiency, which enhances the information available to media buyers and increases ROI. Such a streamlined process enables advertisers to spend their budgets with a higher likelihood of conversion and publishers to earn the most money on their inventory.

A Brief Tour of Online Display Advertising: With exposure to banner ads continuing to rise and Facebook getting into the exchange business, now’s the time to learn a bit about the display industry  (source:


From a technical standpoint, Big Data analytics software vendors such as SQream Technologies are enabling Data Scientists to significantly reduce the query latency and the time spent in model training, ultimately affecting the frequency in which a model can be applied to production real-time bidding.

SQream DB leverages GPU technology to improve the performance of columnar queries by at least 20x on large data sets.  The amazing processing speed on heavy analytical workloads is combined with flexibility and ease-of-use wih standard SQL, and industry standard drivers like JDBC, ODBC, Python and R.


Conclusion: Calling All Data Scientists

Adtech, driven by Big Data technology, is increasingly taking center stage as both publishers and marketers look to propel their organizations forward. Since much of the work is based on driving a competitive advantage, data science plays a crucial role in developing a corporate strategy.

In order to rapidly discover competitive insights, Data Scientists need to have access to advanced analytical tools that can empower them technologically performance-wise, as characterized by high flexibility and ease-of-use. Fortunately, these tools are readily available. As such, today’s data scientists can use their skills rooted in statistics to build models and systems that can facilitate the advertising buying process to reach more of the right customers in different environments. Such a development also has the potential to reduce the short and long term costs for an organization in a substantive way.

In the future, capturing, measuring, organizing and segmenting data is what will separate winners from losers.