Plans & Pricing
Does My Company Subscribe?
Omar TawakolGroup Vice President and General Manager, Oracle Data CloudOracle
As the group vice president and general manager of Oracle Data Cloud, Omar Tawakol oversees all of Oracle’s data management practices and strategies, including the development of its identity graph. Tawakol recently spoke with eMarketer’s Lauren Fisher about the dangers of taking deterministic data at face value, and the increased importance of using first-party data for identifying audiences across devices.
eMarketer: When it comes to cross-device targeting and device-identification methodologies, it’s more evident than ever that in order to get the necessary reach and scale, both deterministic and probabilistic approaches will be required. Do advertisers accept this?
Omar Tawakol: Yes and no. This is what most people are doing for the mix: They’re buying deterministic matches and consider them to be accurate. Then, they realize they don’t have enough reach, so they’re adding probabilistic. But I don’t think that’s the right approach.
The right approach should be taking multiple deterministic matches from multiple places and multiple probabilistic matches from multiple places and scoring the heck out of them so you have one confidence factor across everything. You have a lot of people who have tried to convince the market that deterministic matches are accurate, and that’s not always the case.
We’ve worked with some of the best out there, and we’ve found those sources as high as 50% inaccurate. So you can no longer just accept the distinction that deterministic is for accuracy and probabilistic is for reach. You need to score it all.
eMarketer: Can you share an example of when deterministic data might not be accurate?
Tawakol: One way you get inaccuracies is when someone does a deterministic match on a household and ties it to an email or a cookie, and then the household moves. If someone is going back six or seven months to get big reach, they might be using that old, bad data.
Now, how might you be able to identify that? In our case, we cross reference Datalogix data on 1,500 retailers that shipped goods. So because they shipped goods, their address had to be up to date. And so when we score the matches we’re getting from others against that, we can see errors, even though those sources are deterministic sources.
eMarketer: You mention household data, and I’m wondering whether a lot of companies today are using aggregate household data and extrapolating it to single devices. If that’s the case, do marketers realize this? Or, is there a disconnect?
Tawakol: Absolutely there’s a disconnect. Marketers don’t realize this is happening in most cases. Some sophisticated ones do, and it’s in areas where it is absolutely appropriate to do it. For example, if you look at the consumer packaged goods category, most of those purchases are household purchases. And there are household devices for which this makes sense. An IPTV is one of them, and usually so is a tablet.
It’s appropriate when those things happen. However, if I’m looking at my cellphone and my computer, those are usually personal devices. So the cookie on my computer and my IDFA should be treated as such if I’m capable of understanding that distinction. For example, in my household, it is five ladies and a guy, and we have lots of cellphones. And they should stay distinct.
eMarketer: What are some of the most common things you see clients do in the way of cross-device targeting?
Tawakol: One of the most common things is people take CRM files and onboard them into both online and mobile to match in places like Facebook and Twitter or other walled gardens that have registered users online and in mobile.
I think that’s actually very good. There are a lot of aspects about walled gardens that I don’t like, but this is one that I do like because they have a login in both places, and they’ve shifted most of their usage to mobile. So if you look at Facebook, a lot of its use is already mobile. So when you do these matches in two places like that, you’re getting real people and you’re getting the majority of mobile. We’re pretty happy about that.
You've never experienced research like this.
Nearly all Fortune 500 companies rely on us.
Inquire about corporate subscriptions today.