Sramana Mitra: Let’s switch a little bit to a different line of questioning. If you look at this approach of AI in the publishing and advertising space, what are the trends that you’re seeing? I’m specifically interested, however, in understanding where AI is making an impact and what kinds of trends have you seen in some of your industry peers doing what they’re doing using AI.
Patrick Shea: I think it’s specific to a lot of industries but it’s very prevalent in the digital marketing ecosystem. One of the biggest trend is the level of complexity and the fact that there are so many point solutions that are trying to solve a very specific problem. That’s a good thing. The barriers to entry in this market are lower than in others. It allows people to compete on value. When they provide a different value, they insert themselves into the larger stack. However as that happens, that level of complexity can create friction between buyers and sellers of digital media. As this industry continues to evolve, there will be more open platforms. There’s going to be more common standards that really don’t exist right now.
At the current state of affairs, most of these systems don’t talk to each other in an extremely efficient manner. You have players like Google and Facebook that are very heavily invested in their own stacks. There are other players in the space that are trying to create a more open ecosystem. I think that the trend towards standardization and communication between platforms is essential to allowing the automated decision-making and the algorithmically-driven marketing that offers some of the promise to really take hold and achieve the scale of success that we know we can offer.
Kevin O’Malley: Another big trend, especially with AI, is there’s a few companies who are successfully using AI to drive marketing decisions. We see it often on the marketing automation side. We really think that ad tech and marketing tech are going to continue to blur. We’ve seen that enterprise marketing companies are showing great interest in automated display advertising whether that be display banners on desktop or mobile, as well as video in Facebook.
Traditionally, enterprise marketing companies have just focused on email. I think they are starting to see the value in display advertising as well. Traditionally, AI has been focused on that marketing side. I now think there are a few companies in the space that are doing some interesting things on the display advertising side. However to Patrick’s point earlier, we found that artificial intelligence requires tens if not hundreds of millions of impressions to gather that data and to actually find some learning from that data that can apply to the advertising campaigns. The vast majority of campaigns – 99% of the advertising campaigns out there – are probably less than 10 million impressions. There’s just not enough amount of data and information for these black box algorithms to have a really profound effect on advertising.
We took a step back and we built, what we call, AdDaptive Intelligence. It’s a combination of machine learning and Big Data with specific human domain knowledge. You really have to point the algorithms where they’re going to go. If you just have a black box algorithm, you’re really not effectively driving good decision in advertising.