Xaxis Hatches Three-Prong Plan To Improve Programmatic Buying

nicolle-pangis-xaxisXaxis is placing three bets on how it can evolve programmatic buying, according to global COO Nicolle Pangis.

One, it wants to understand the entire consumer life cycle. That means using historical data and a team of data scientists to figure out when someone enters the market for an item and the best time to influence that purchase.

Two, Xaxis wants to inject its own tech and algorithms into existing demand-side platforms (DSPs) through an initiative called “Co-Pilot.” It started with AppNexus, in which WPP owns a stake, but is open to all DSPs, though not all are equipped to let an outsider’s algorithm and data management platform (DMP) take over. Other machine-learning-focused initiatives will follow.

With ad blocking on the rise, Xaxis’s final area of attention is the ad creative experience. It wants to optimize poorly performing creative in days, not a month, and improve the relevancy of its messages.

Pangis, who began leading M&A and the Xaxis product, engineering and data science teams in January when she assumed the global COO role, must implement all of these changes.

She started out at 24/7 Media in 1999, and except for a four-and-a-half-year stint at Cadent she’s been part of the company ever since. She stayed after WPP bought the company and made it part of its media innovation group, and then took on the CRO role when 24/7 merged into Xaxis in 2014.

At the DMEXCO conference in Cologne, Germany, last week, Pangis talked to AdExchanger about what to expect from Xaxis and how it’s thinking about header bidding and the demise of cookies.

AdExchanger: How is Xaxis thinking about activating data now versus a few years ago?

NICOLLE PANGIS: By 2020, the expectation is there will be 1.7 megabytes of data per second per human on the planet. With that, the question is what data is relevant for what type of campaigns? From a technical perspective, we are concentrating on the activation of data in a smart way. You have to sift through and find the right data.

What other advertising problems are you trying to solve?

Through our data activation platform, Turbine, we are surfacing audiences that are most relevant for the campaign. If you can look back in time and see the life cycle of the consumer that converted today, and see what they did 30 days ago, 20 days ago and the few days leading up to the action, you can model against it.

What’s the difference between Xaxis and GroupM Connect?

GroupM Connect is a service for clients – programmatic buying on behalf of clients. Xaxis is a programmatic audience company. Everything we do is data-driven based on its audience. When we work with a client, we are offering products, not services. We have a data platform called Turbine, and we have artificial intelligence called Co-Pilot which takes data from Turbine and runs our own decisioning into the DSP. We show the campaign and audience data through our visualization platform called Spotlight.

What happens when consumers move to cookieless environments like mobile web?

We are moving into a cookieless world. The combination of Turbine and Co-Pilot allows us to leverage a single consumer ID across all platforms. [That] allows us to ensure we are not targeting the same users across media and ensure we can sequence messages properly across the consumer journey wherever we see the consumer.

As you refine ad targeting, is ad creative multiplying to address these moments?

In 2005, when I sold campaigns, I would get three creatives, and they all looked pretty much the same. There is more flexibility now. Will there be a billion ad creatives? I don’t think it’s a bad thing to have more versions of ad creatives, as long as the brand value is captured in all the versions of the creative, and it’s benefiting the consumer experience.

Xaxis works directly with many publishers, and that mode of buying appears to be back in vogue as many shift from buying programmatically across the open web to buying on a curated list of sites. Why work direct with publishers?

We use platforms to access inventory programmatically, but that doesn’t mean we don’t have relationships with the end point – the publisher – to drive best value and results for our clients. We have dramatically cut down on the number of partners because we found that having fewer, stronger partnerships drives better results. You can get a lot more strategic on the publisher about placements that are working, and how you are going to access different types of inventory, so it’s been a great strategy.

How is Xaxis looking at header bidding? Would you switch to integrating through the header versus the tag-based approach you have now?

It feels like in our industry, it’s this or that. My view is that this is not an “or” discussion. You will have tag-based [buying], PMPs through DSPs. We are in 40 different markets and nothing is the same in each market. They are at different maturity levels. Our job is to deliver campaign results for clients, and we are going to make sure we do that on quality inventory with the publishers we work with.

What’s the value of a publisher’s data – audience segments they create – versus an advertiser’s data?

Any data set would end up in Turbine, our data activation platform. We look across the walled gardens with a single language, which is Turbine. Publisher data, it’s one piece of the puzzle in hundreds of pieces of data we have. What we wouldn’t want to do is use different data sets everywhere we are targeting because it’s difficult to bring that in and create the appropriate analytics, and everyone has different definitions [of audience segments].

This interview has been condensed and edited.



Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!

1 Comment

  1. Not sure if Nicolle is being literal but her example of 20 – 30 days look back on consumer behavior – at least for considered purchase products – is a very small sample. Given the non-linearity of the customer journey, it would take different timelines based on the product to gain an understanding of what success looks like. The concept is correct – the sheer amount of data that will be sifted across time will be formidable – given our own experience.