AdExchanger: How did your relationship with Facebook begin?
ERIC ROZA: Sheryl Sandberg and I first discussed the possibility of working together three and a half years ago. I was introduced by Sheryl to Brad Smallwood, the head of measurement and analytics.
How did the first year of offline conversion tracking go?
Intense…a lot of groundbreaking analytical work. No one has ever measured media at this scale. We're going to take the majority of impressions that get served and measure those back to in-store purchasing.
There's two things you get with that. You get confidence. When you have small samples from a panel of consumers, you can have bias from those samples. As you get a large universe, the biases go away and you get more confidence in the numbers.
Also, you can derive real insights. You start to look at how different sub-populations respond, and those become more actionable. Whereas historically people were almost giving a yes or no answer – this campaign worked or it didn't – it becomes much more interesting to say these parts of the campaign worked, these parts didn't, but next time we can do X and Y because of it.
Did you have visibility into how Facebook's advertisers were using the data that was coming in through the initial partnership?
Datalogix actually did the analysis. We almost have two roles.
One is we're an abstraction layer. Our job is to protect Facebook's impression data and the retailer's sales data so that no one gets access. You're almost assisting as a systems integrator and abstraction layer to keep the data safe and partitioned.
The other value is doing rigorous analytics. A really challenging problem is proving that an online or mobile ad actually caused more toothpaste or more cars to be sold. That's because – as we all know – display ads are pretty small.
Brad and Sheryl foresaw that for Facebook to reach its potential as an advertising medium, we have to be proactive about proving this stuff works and then improving it. Maybe we find out it doesn't always work because – let's face it – there's no advertising that always works, but we need to know so that we can keep making it better.
Is Facebook more compelling than traditional display, since it's viewable and arguably fraud-proof?
Most Facebook advertising works very well, and we've now measured over 100 campaigns.
It's an environment that's very well controlled by Facebook. You can derive insights and then act on them at a scale that hasn't been possible on the open web for non-direct response. For direct response, there's obviously been an ability to optimize. For the traditional brand advertising, it just hasn't been there.
We've even done some tests comparing Facebook to some offline ads. I think you'll see more of that in the coming year.
Before we get to Custom Audiences and targeting, how has the measurement piece worked out for you from a business standpoint?
In the year and a half that we've put into it, we've just put thousands of man-hours into R&D. Our focus has been less on short-term revenue – although short-term revenue has been nice – and more on building this foundational platform. We call it DLX ROI.
Let's go on to the more recent stuff, which is your matching of CRM and purchase data to Facebook audiences for ad targeting through the Custom Audiences program. How did this come about?
Facebook started to realize that if we could bring in signals around what people have purchased, especially in categories like consumer products and automotive, that could be a powerful complement to this great data set we have already.
That dialogue started in earnest about six months after we kicked off the DLX ROI project. We had developed trust in each other as organizations. We were able to say "Here's another use case" versus "Let's start from scratch." We had this platform and this relationship that could be leveraged, not to mention the hundreds of folks working with each other at both companies by that point. It put us on a timetable that was much more aggressive than you might have seen otherwise.
Within a little more than a year of our measurement pilot, we had started the Custom Audiences pilot with Facebook.
After the Custom Audiences pilot, Facebook added partnerships with Acxiom, Epsilon, and BlueKai.
Some of the other guys they're working with host databases – which isn't a business that Datalogix is in, by the way – for a lot of the Fortune 500 brands. It made sense for Facebook to work with the companies hosting the databases and bring in that data.
Much of our work was on building out robust segmentation for brands who didn't have their own data or needed to supplement their own data. The majority of the partner categories that have launched in Facebook are from Datalogix – more than two-thirds of them. Obviously, the other Facebook partners are companies we know well, partner with and respect. We each have a role to play here.
We've had an exciting opportunity to get to know Facebook's analytics team, the sales teams, the vertical category teams. We've gotten to know some of the big Facebook Preferred Marketing Developers but haven't been able to work with them yet because the DLX ROI work has not yet opened up to the PMD channel.
We've been doing this kind of stuff on the open web for years. What's really nice for us is we can now give our big agency and brand relationships some consistency.
Stepping away from Facebook, do you see an opportunity in similar relationships with other social media and native ad platforms?
Yes, we do. We haven't announced a lot of stuff with some of the other big publishers in the social world and on other native ad platforms, as you put it.
We are working with most of them at this point. It's kind of in a test phase, and I think everybody wants to make sure it's going to work and add value. Some relationships that have been talked about publicly [are] Yahoo, Google, Aol, and within the last year eBay as well.
Is that on the measurement side or targeting side?
Both. With some, it's been led with one and bled into the other. We think of ourselves as a platform that has this great ability to connect offline data with digital media. That's what we do really well.
It's almost like an infrastructure question. We've been working with the Genome guys at Yahoo for years, well before they were a part of Yahoo and were Interclick. When we started working together, it was all about finding the right audiences. That was differentiating. Over time, the relationship expanded into measuring. Did those audiences buy more offline, especially in areas like consumer package goods and automotive?
Yes, either one can lead depending on what the use case is.
Are you guys profitable?
We have some parts of our business that are very profitable and some parts that are less so. Because we have businesses at various stages, it's almost like a portfolio. We have a burgeoning international group. We have the measurement products, which are younger, so those are in an investment stage. We have an offline targeting business that's very profitable. It does direct mail analytics.
How does the Facebook piece compare to the rest of your business?
Facebook's become a very important partner of ours. We're different from a lot of people in the Facebook ecosystem in that the vast majority of our business is not Facebook.
Facebook is obviously such a big presence that a lot of companies end up being in orbit around Facebook. For us, Facebook is one of our most strategic partnerships. They're pushing us to improve the product.
The Facebook relationship has brought a lot of press coverage. What's been the impact there?
It's been positive overall, but it's not unmitigated. At times it feels like there is a mainstream press bias against big data. That's hard and feels unfair. We know we're good people doing important work, and we not only do it well for our customers who are publishers and brands and agencies; we think we also do it with a very high degree of integrity.
That is going to be an externality of working with Facebook. We've chosen at this point to embrace it. We've tried to be transparent about what we do. We're very proud of the role we're playing. I think it's a thrilling time, if you're a data person.
It becomes a privilege to be part of telling the story. You just have to be aware that not everybody comes in with a neutral perception.