AdExchanger: How does the Oracle Data Cloud coexist with the Oracle Marketing Cloud?
OMAR TAWAKOL: The Oracle Marketing Cloud is a software stack where you license technologies for execution. Data Cloud is a data-as-a-service provider independent of the software stack. It provides this ability to acquire second or third-party data and plug it in anywhere you want, including the OMC stack.
Where does the data end up?
OT: A huge amount gets plugged into Google, Facebook and Twitter – all media platforms a customer works with. Think of the Marketing Cloud as a vertically integrated software stack and the Data Cloud is this horizontal data access you can plug into any other stack.
OT: BlueKai came with this strength in “anonymous” and had tremendous reach, but we were really missing the complementary components that Datalogix brought. We saw a future [where the] web was half open and half closed [and where] several players who would focus on real identity matching.
That kind of anonymity wouldn’t be enough from a data standpoint. Then, you’d need to be able to [merge] these channels through known identity data and transaction data. Datalogix was the world’s best at doing those two and applying the data science to judge the quality of data, pull all of these facts about an ID together and come up with a confidence probability factor.
The problem in the industry is … [many] people say they have an ID graph, when they only have individual pieces of that. If I’m a marketer, I need you to connect all of it– whether it’s known or anonymous, online or offline, deterministic or probabilistic, user level or household. Our goal is to unite all these pieces of the ID graph for the customer, and this is a multi-year, very big investment project to get right.
Where is there work to be done?
ERIC ROZA: Additional layers of complexity arise when a new consumer platform gets broad-based adoption like a Pinterest and Snapchat. Those are new identities on top of everything else. [Marketers] want to create a consistent experience for their customer with some level of coherence across touch points. In the mobile ID world, it gets even more complicated and you have people claiming things that look like an ID graph purely based on unvalidated IP addresses or bad location data, but – candidly – getting it completely wrong.
We need to do work in almost every area still. We’re quite a bit further along than other companies. However, where we are versus the ideal end state – [where] a big marketer can across all their media touch points hit Jane Doe 10 times per week [at certain hours with a consistent message]. It will be a while to get to that idyllic state. But you don’t need to be there fully to provide value. You need to keep getting 10, 20, and 30% better every year.
What did Neustar’s acquisition of MarketShare say to you about further DMP, marketing analytics and attribution mergers?
OT: There is this debate [about whether] the DMP stack and attribution stack need to be collapsed. Probably yes, but in the short run we felt there was no concentrated share of market for any one attribution vendor because there were so many of them and none of them had critical mass. If we [sold] our DMP [and] attribution together, [it’s likely] most clients wouldn’t want our particular selection of the stack. They would want it to be integrated with everybody. This is one reason why the Data Cloud was left open. We’re not forcing people to buy data through the Marketing Cloud.
How is Datalogix different from a multitouch attribution system?
ER: [Multitouch attribution systems] don’t really [determine] whether a digital campaign drove offline sales, which is an incredibly complex problem. We spent four years getting the big CPGs and automotive and retail brands comfortable with our methodologies linking digital exposures to offline incrementality. An MTA platform might say, “You get 100 online conversions and ‘X’ amount of them were due to paid search or display.”
Attribution is very different than what has been our stock and trade with Datalogix ROI measurement, which is, “I ran an ad campaign and because people took a lot of actions offline and on other channels, I can’t assume that $100 transacted offline was due to my ad campaign.” We help advertisers determine, “Of that $100, would $50 or $95 have been transacted regardless of the ad campaign?”
What are your thoughts on TV and programmatic targeting?
OT: If you can at least individually measure that a person saw an ad and then made a transaction, you can start to change the way you allocate funds to TV. This will drastically address the huge part of the TV market that is accessible versus waiting for each impression to be made targetable. Increasingly, the partners we work with have the ability to target down to the impression and eventually it will be near 100% of the TV market that's targetable.
What’s an immediate focus for Data Cloud?
OT: [We are] focusing on what consumers do, buy or say. A fast emerging area is where they go. To know you saw an ad and showed up later at a car dealership or competitive dealership. Another area is bringing more unique mobile assets in addition to online-to-offline measurements. Also, we’re focusing on international data assets. Oracle is strong in the US/UK, but increasingly people who have a DMP need third party data from around the globe.
This interview has been edited for clarity and length.