AVI SPIVACK: The whole point is to unlock audiences across a spectrum of categories, and if you want to expand your audience, you need to tie media to categories without having to scale across dozens of retailers.
There’s still some skepticism in the industry about co-ops. Do you feel that?
We do hear from advertisers that they want to keep all of their data behind their walled gardens to use just for themselves, but that’s a concern that’s related more to the security and privacy side of the equation than anything else.
In terms of the much larger retailers, we’ll hear things like, “I have hundreds of millions of uniques a month – why would I want to contribute all of that data into a co-op?” For big advertisers like top Internet Retailer category leaders, it may not make sense to contribute because they’ll be able to gain enough scale on their own.
Who does it make sense for?
The key question that any commerce company has today is this: In such an unbelievably competitive landscape, where companies like Walmart and Amazon exist, how can I compete and continue to evolve? It’s a struggle out there, especially if you’re focused on the retail side.
How can you inform other marketing channels like search or site personalization without using a wider data set? Who buys what and how much? With what frequency? The co-op is about giving the retailer the ability to understand that.
Data co-ops try to answer the question, “What if you knew more?”
So, basically, a bit like what happens when catalogers share mailing lists in the offline DM world?
Offline co-ops have existed for 25 years. Early on, catalogers saw that they couldn’t just use their own data because it would be a limited pool, so they came up with a way to acquire new targets.
It all comes back to data collection. Yes, you can collect from your existing customer base, but that will only have so much effectiveness before you’ll ultimately need to figure out a way to extend it.
I had a conversation with a home goods store recently which said it knew who 50% of its web visitors were because they’d either been cookied or they’d logged in. But the other half? They didn’t know anything about them. You can’t really create a personalized site experience with different versions of creative unless you have an extension out in the commerce space and understand what else your users are doing out there. That’s the larger business question.
What about the privacy implications?
There is a massive dissonance going on. There is so much concern around online privacy, yet we’re not collecting a single shred or element of PII. We know nothing other than behavioral attributes. We’re ignorant of specific consumers.
There is that creep factor when someone sees an ad following them around the Internet. But is that really worse than buying something from Pottery Barn and then suddenly receiving a Crate & Barrel catalog at your home? In that case, they have your name and address.
But whether one is “worse” than the other doesn’t mean there aren’t issues with both.
It’s also about choice and creating an open system where consumers create their own profiles, which are then used to inform more personalized Internet experiences overall. Ads won’t go away. In order for the Internet to be free, there needs to be advertising. Our goal is to use data and technology to make that experience more personalized and safe. I think that’s something people would actually want.
What kind of cross-device data do you have?
We see between 350 million and 500 million unique shopper profiles across phone, tablet and laptop. If a person has one of each, they’re represented three times. The data is not deduplicated. But we are using our technology to dedupe these users and get it down to somewhere like 140 [million] to 160 million truly unique customers.
What’s been happening since the Arcametrics acquisition, the predictive modeling company Adroit acquired in November?
Before, we relied on MediaMath assets, but now we can be more independent, although still work in conjunction with the MediaMath platform. Our goal with the acquisition was to differentiate Adroit both from a data asset product standpoint and technologically. Arcametrics had been a partner to many DSPs in the space, including MediaMath.
Since November we’ve made strides in improving and overhauling the modeling platform. We brought on their team and added three new divisions headed by three of their core guys: an engineering team, a data science group and a more robust product development group. We’re also in the process of hiring a full technology team.
We’re looking to establish a predictive modeling product that’s directly tied to the data in the co-op. Arcametrics will have a big hand in that.