We are not replacing Google’s AI, we are augmenting it: Sara Robertson, Xaxis
Robertson, VP Product Engineering - Xaxis, spoke about the unique solutions that Co-Pilot offers, the scope of programmatic in India and the need for making AI more atuned to contextual advertising
Published - Mar 20, 2018 9:00 AM Updated: Mar 20, 2018 9:00 AM
With programmatic becoming the cornerstone of media planning, new ad tech products are constantly redefining programmatic practices. As part of this technological evolution, GroupM owned Xaxis has been pioneering new technology for the advertising industry to offer simplicity and accountability to clients. One of its popular products—Co-Pilot is making news for offering solutions to many programmatic bottlenecks and drive measurable outcomes for clients.
In an exclusive interview with exchange4media, Sara Robertson, VP Product Engineering - Xaxis, spoke about the unique solutions that Co-Pilot offers, the scope of programmatic in India and the need for making AI more atuned to contextual advertising. Excerpts:
How is Co-Pilot different from other existing programmatic products?
If we compare it with existing platforms, Google has incredible Artificial Intelligence (AI) offerings and all of the major DSP platforms have a similar engine. But their goal is to build a generic solution. They solve 80% of the problems for 80% of the people. They are trying to target industry standard outcomes and make them work for everyone. On the other hand, what Co-Pilot does is to add extra layer of specificity on top of the platforms to try to close that 20% gap. So if you are looking for an outcome specific to your business, it’s not probably present in one of these default platforms. With Co-Pilot, we can use machine learning and optimize it to gather insights and then push those insights into these platforms. So we are not replacing Google’s AI, we are augmenting it to try to drive it into a customized direction for benefit of the clients.
Have publishers become more open to sharing their premium inventory on programmatic?
As we know, the majority of digital ad spends are going towards programmatic. Every publisher has yield optimization problems, and they want to sell as much of the inventory as they can on premium price, which is why header bidding became a big issue and why people negotiate for first look and guaranteed buys. I think most publishers have found that the value of selling your entire inventory is greater than getting a premium price on some of your inventory and then having a ton of the remnant that you sell for pennies. So I think its super important to make sure that we are helping publishers monetize and get the most value they can because they are the internet.
Does India hold a big promise as far as programmatic is concerned? What kind of unique challenges does this market offer and how are you trying to deal with it?
Indian market is exciting in all industries. With rapid development and proliferation of mobile devices a large number of population is coming online. In terms of growth metrics, India is on top of the charts. Now the question is-- what can we do to take advantage of that, and secondly; how to facilitate and empower that growth. In an emerging market where people are just discovering internet, we don’t want to destroy their experience by burdening them with ads. So it’s a delicate balance and there are some specific challenges in India in terms of the diversity of languages. So a dynamic creative solution needs to be put in place so that we can effectively serve in such a diverse market, and we are working towards it.
How would you define the future of media planning?
I think in couple of years you won’t be able to get away with not being data driven. Everything will be data driven and your competitors will be using data for insights and making recommendations. Like planning specifically is not going to be based on intuition anymore, it will be driven by facts. I think the skill sets required will be very analytical and we might move away from needing code because systems will start to get created which can manipulate data visually.
Can AI match the emotional and contextual aspect of human thinking in the coming time?
I have not tried to solve that problem with AI yet, but it has been mentioned so many times that I feel we need to find a solution to this. I think AI is capable of being a tool that can be atuned to any given situation. But it will just remain a tool that humans wield until the planners, the traders, the clients, the consumers and the engineers decide to focus on the contextual aspect of it and solve the problem of not retargeting same things forever. I think over time systems and technology will mature so that we don’t have to bombard consumers with messages without understanding the context of that communication.
WhatsApp, Instagram, LinkedIn, Twitter, Facebook & Youtube