YOptima launched DeepView, an AI driven recommendation engine to address a problem popularly known as ‘Simpson’s paradox’ as applied to Programmatic campaign analysis. With DeepView in its current form, YOptima aims to streamline its own campaign management function improving its capacity and effectiveness by at least 30%. In the next version, DeepView will be offered as a tool which can enhance campaign management with its advanced data analytics and intelligent recommendation engine.
Simpson’s paradox is a well known occurrence in probability and statistics in various data problems, where patterns appear in a group of data but disappear when these groups get combined in a larger group, or vice versa. Using data without the context of underlying behaviour can lead to misplaced decisions. For example, analysing some campaigns indicated that mobile ads were not performing well overall. Digging deeper, we saw that only ads on a few websites were not performing well and because of their higher volume were bringing the overall performance down. As opposed to stopping all mobile campaigns, we instead removed the erring websites and achieved an appreciable overall improvement in campaign performance. This may have been an easy problem to solve without the help of mathematical tools, but with increasing dimensions and data set sizes, it can get exponentially more challenging. DeepView addresses these problems and makes life easier for campaign managers.
Satish Kadu, the CEO explained, “DeepView employs AI and Statistical techniques to analyse data for many different possibilities and combinations which can yield the best outcomes for the campaign, much like Dr. Strange does in ‘Avengers Infinity War’ to determine the Avengers’ chances of success against Thanos. A simpler analogy would be what Google Maps does when it analyses traffic data before recommending a suitable route for you to take back home from work. DeepView essentially examines these options and goes several dimensions deeper and would in an equivalent recommendation tell you not to take a particular route because you are driving an SUV which cannot make it through narrow streets, while it would be perfect for a two-wheeler.”
For a campaign manager, DeepView processes the data for patterns and recommends a few intelligent courses of action, which saves her several hours of intense data crunching. It essentially harnesses technology to hasten the process of intense data crunching which otherwise would take humans at least 3-4 times longer. What this does is reduces the time to reach the optimal combination of campaign parameters or the optimal ‘y’. It also can examine more possibilities than a human can given constraints on time for a human employee. However the critical element of creativity and judgement are left to the campaign manager, which we believe is the way it should be.
AI and other technology are great enablers for humans to take data driven decisions, which is where we see YOptima really taking the lead in creating tools and solutions for the AdTech industry which will change the way we all work.