• by Tobi Elkin  @tobielkin, April 6, 2016

It's a goal for many industry stakeholders: the ability to use consumer data -- and insights drawn from that data--to inform creative messaging. How does that happen?

Michael Davis, head of creative for Conversant, explains the media company’s approach. Conversant offers video, display, mobile and CRM programs, and sees Google Ad Network, Yahoo Audience Network and Facebook as its main competitors in terms of its focus on one-to-one personalization and relationships with both brands and agencies.

Real-Time Daily: What is the opportunity for data to inform creative?

Michael Davis: On one side you have creative agencies — and most marketing companies on the agency side, which want data-driven creative. They need data to give them the answers on how to build creative stories. That’s the agency model.

While the industry is inundated by verification based on clicks, we can’t verify clicks — there’s too much ad fraud. We verify what real people are actually doing. We get the data from agencies and brands — and then on our side, we help them break the data down into thousands of different pieces.

On the media side of the business that we work on, the data actually informs decisions, not creative. We work with lots of creative and media agencies. Based on the 160 million consumer identities that we have, there’s a creative algorithm that will show you all of the data that informs the decision on what people should actually see. It’s advanced personalization.

Real-Time Daily: How does that work?

Davis: Say you have Brand A, and on that brand there might be three campaigns running. We might create seven different looks for those three campaigns. We work with agencies or brands to help them with those treatments and how they should look across devices.

Of those seven different treatments, we might create 30 to 50 different headlines and 50 different opportunities.

For example, based on weather and geo-location, if someone is buying shoes in Florida, we won’t send an image of snow boots.

We might have 80 different product images, headlines, calls to action, backgrounds, copy and text.

In the algorithm, you have the potential for 56 million potential iterations on one campaign. They’re all multiplied in the algorithm, along with 7,000 attributes (identifying people by gender, where and when they’ve shopped, what they’ve bought, etc.) and we have data on 160 million individuals.

The holy grail is, how many real people can we talk to today? We have 160 million real people across devices—we can talk to that many people.

The engine we have makes decisions on who the right person is, where they’ve been in the last hour, two hours, 10 days, in-store, etc. We have online and offline data that’s anonymized. 

Real-Time Daily: Are you doing retargeting?

Davis: We’re not doing retargeting. We’re using first-party cookies and work with online and offline data.

Real-Time Daily: How are you dealing with ad fraud and non-human traffic within this system?

Davis: There’s a difference between real people vs. bots and clicks. We go through extensive data scrubbing to ensure our ad quality.

For the ad inventory, we’re up to 96% accuracy and cross-device matching verification. What that means is that we have a way to talk to real people who are doing real things because our advertisers provide us with their offline and online data. That data enables us to verify second by second whether those people are real, where they’re buying and what they’re doing.

Real-Time Daily: Can you walk us through an example?

Davis: Take the hunting, fishing and camping gear retailer Cabela's, for example. We want to talk to a real or potential Cabela's customer.

We’re optimizing creative right at the time when they’re online, deciding what layers to send.

We assess your personality and lifestyle attributes and where you are cross-device. One Cabela's ad might have 10 different headlines. All of those decisions come together. The ad isn’t pre-rendered — it gets put together at the second the customer goes to the site.

We offer personalized display advertising that’s rendered at the time of the bid. Within 40 milliseconds, we’ll make a decision on what ad to serve you, so it’s a truly one-to-one transaction based on who you are and your attributes. We might send you any one of 56 million iterations of an ad, offer or promotion.

This isn’t just about putting a new headline on the ad. We put all the layers together within 40 milliseconds at the time of the auction, in real-time.

With the CRM data we have, we can understand anonymously where people are going.

We have 100 people in the creative department. The decision on what creative units get rendered is constantly changing based on the analytics. The analytics team works within the engine to make the decision.