AdExchanger: Does Annalect mostly service the other agencies within Omnicom, or do brands have direct relationships with Annalect?
CHARLES BUTLER: Predominately it’s the agencies. Right now, [direct-to-brand relationships] are the minority case, but there are plenty of situations where the relationship is direct based on specific analytics a client might want.
What sort of requirements would a brand client need that makes them go direct to Annalect?
In the short time I’ve been here, it seems to be around a specific type of insight outside of the original contract or scope at the agency level.
So if the agency can’t get a certain data asset, the brand might go to Annalect.
Right. From what I’ve seen so far, it seems to be a derivative of some other element of a campaign.
How do Omnicom agencies engage with Annalect? Is there an Annalect analyst running around different departments, or is it like Commissioner Gordon calling up Batman?
It’s more like the former. [Annalect is] mostly baked into a relationship at the beginning. Secondarily, it becomes an add-on or a cross-sell. That’s sort of like the Commissioner Gordon thing – show the Bat Signal.
Is Annalect a service or a technology?
It’s both, but it depends [how you look at it]. From the Omnicom family, it’s more a technology and less of a service. From the client side, it probably looks a lot more like a service.
What’s in the Annalect stack?
I’d love to be able to [say], but being this new, I have not had a chance to comb through it myself. Here’s what I have seen, and it’s completely nonexhaustive: From the DMP side, it is the Neustar-Aggregate Knowledge combination of data sources. The technology that’s supporting that is a combination of commercial software pieces like Amazon RedShift, which are informatic databases, and a series of analytics tools – some of which are commercial, some of which are in-house created – and a series of visualization tools as well.
It’s a gamut of technologies that are both commercial and also proprietarily developed.
Where are you right now in building the Annalect platform and where will you be in the next six months?
At this stage, we’re midway to two-thirds of the way through that mission. In six months we want to be fully realized. However it’s a continuous evolution.
We already have a backlog of new features clients are asking for that takes us into the beginning of 2016.
How will those new features connect with the platform? Will they be baked in or will there be something like an Annalect application shop?
We want to put into the platform the features and capabilities that will be meaningful across the suite of clients, and not just create a bunch of one-offs.
Clients frequently just need to be steered through education. Often, they ask for [a feature] because they don’t know another way [of doing things]. Showing them a new direction, with the right sort of hand-holding, is probably the combination we want.
But for [the platform] to be successful, we’ll have to [account for] the evolution of features that make sense at a platform level, but are also market differentiators. We’ll also need to accommodate [future] opportunities, meaning we’ll need to be able to adapt to the next technologies and the meaningful data sources they’ll generate.
How do you build for that adaptability? Through API plugins?
Generally speaking an API is the right strategy, yes. But that’s not exhaustive. Think of the API as a way to ingest multiple different data sources and keep your options open. That’s why the future will be driven by CRM-type data entries. The clients are very smart and rich in terms of their customer data. That’s useful to them and to us, as an analytics shop providing solutions for them.
The platform [needs to be] extensible. We need to keep the architecture elegant and simplified, but built in a way we can add new code and features to it.
Were you part of the team that developed the AOL ONE platform?
Yes. It was in the capacity at AOL Platforms. I was on the technology operations side, so I’m very familiar with what the ONE platform is all about.
Is it similar to what you’re trying to do at Annalect, or does that analogy not hold?
It’s similar, but it doesn’t hold. The difference with AOL is there were a handful of very dissimilar vertical technologies. ONE is trying to create a tiered level unification that presents itself as one to the outside, but will actually be different technological venues underneath. In a couple of years from now, it should ideally be far more integrated.
That’s a contrast to the Annalect world, in that there’s not all these different technologies driving the business. The basis of the [Annalect] platform is unifying it from the ground up, instead of from the side, mashing technologies together.
So AOL had a bunch of different tech and stuck it all together. With Annalect, you’re building on top of a common platform?
[Annalect] had the good fortune of having the concept first and not making the mistake of creating bunches of vertical technologies and trying to integrate them horizontally.
I don’t want to speak ill of my former place. The reason [AOL's] technologies were vertical was because they came via acquisition. The acquisition of Adap.tv last summer, which is the video side of [AOL’s] business, had a radically different architecture from the network ad-serving side and the DSP engine side. So there was sideways integration work.
We alluded to your partnership with Neustar. What role does the data-management platform play within Annalect?
Aggregate Knowledge, the DMP that Neustar acquired, was one of the market leaders. Their capabilities were hugely attractive. And Neustar, given the market they were in, has CRM data [separate] from the AK offering. But there’s synergy between the two. Think of the DMP capability and Neustar with its own CRM data. It is an attractive data set.
I can tell you AOL had a deep interest in Aggregate Knowledge.
I thought the data set and the platform were two separate entities with Neustar. Is Annalect plugged into both?
ADAM GITLIN [Editor's note: Annalect's global managing director of its data group answered this question via email after the initial interview with Butler was conducted]: What we use from Neustar for the DMP is separate from the data selling business they operate. Omnicom has partnered with Neustar for the DMP ... separate from that platform, we may buy from the data exchange just as we buy from other data providers.
What data does Annalect tap? Do you have a data pool, or do you isolate an agency’s data so it doesn’t mingle?
Can I get a high level overview of the different data sources Annalect uses?
BUTLER: We’re focusing heavily on the CRM data. Having a platform able to ingest it is a huge opportunity. The advertisers we have out there are evermore savvy about what they have and collecting more and more. I think that’s the area I would point to.
Is this focus on first-party CRM data a new practice for Annalect?
I’m guessing it’s probably a newer direction.
Have you followed Google’s crackdown on DMPs?
Short answer is no, I haven’t. Being in the third week here and pivoting against the marketing side of the industry is a real change for me. Coming from the ad tech side of the world, we had third-party targeting data and we had the data we were serving and moving in the ads. The DMP was not a critical function for us in the ad tech world.
What different considerations do you need to account for when working with first-party vs. third-party data?
You have to approach it with a great deal of sensitivity and being a trusted steward of conventional data. That’s top-of-mind for us. That situation isn’t technologically hard to do, but requires the right process, rigor and procedures, especially to do it at scale.
In an ad tech world, the data is not PII data. It’s already abstracted.