AdExchanger: You mentioned there were certain backgrounds and skill sets you needed. What were those?
BOB RUPCZYNSKI: To speak generally: Different. It sounds dumb, but you can’t have the same cookie cutter type media planners and media strategists that have come up the funnel and done the same things with the same experiences and the same outcomes. It was just getting different backgrounds and opinions and viewpoints.
[Ad tech company] Turn’s bringing expertise from other industries. Starcom has a plethora of other companies they work with, approaches they’ve done and know about things that don’t work as well. Those vantage points are valuable, which is why the sum of the three is greater than if we were to just stack on top of each other.
How long did the transformation take?
BR: That transition was exactly what we expected it to be. It was a process. There was some good right away off the bat. There were some things we needed to work on over a period of time. It was not: “Hey, six months we’re perfect, we’re done.” I’d say it’s an evolution. We made a tremendous amount of progress very quickly. That last 20% was [harder].
TRACEY PAULL: The other key, instead of having Bob give his failure-is-not-an-option speech to client SVPs, Bob gave it to the whole team. From the associates to the EVPs of the account, everyone felt collectively responsible for delivering the product.
What about Turn’s tech appealed to you? Had Kraft worked with a DMP or DSP before?
TP: They hadn’t worked with a DMP before. They’d worked with other DSPs fused with DMPs. Turn had incredibly strong technology in the DMP space.
We just wanted to make sure we got the DMP side right – that it [could harness] first-party data. [Turn’s] data vision goes beyond media as well. [Kraft was] looking for a technology partner ready and apt to plug into other parts of the marketing organization that wasn’t media. And that’s something Turn could provide.
Was one of the requirements a coupled DMP-DSP?
BR: It wasn’t a requirement. It’s where we ended up.
TP: Frankly, it’s been great. We focused on one partner and any discussion, we can toggle between the DMP and DSP. It’s that fast, rather than: “Hold on, I need to talk to the DMP to see if that’s possible.”
In your eyes, how do DMPs differentiate?
TP: It isn’t one thing works better than the other. It was willingness to adapt the technology to what we wanted. There are things [Turn is] building into their DMP specific to our needs. We have always said our KPI for Turn isn’t “Is it right for today?” It’s “Do they understand what needs to be done?”
Does Kraft have more than one DSP relationship?
TP: Currently it’s just Turn. We have explored some other video relationships and Turn knows that. But right now we’re tremendously satisfied with what we’re doing at Turn.
What’s the division of labor between Kraft and Starcom in this new media model?
TP: Bob’s job is to set the bar. Our job is to deal with all the minutia and details to get to that bar. Unlike most clients, Kraft’s CMO [Deanie Elsner] came into the office. We sat with her and showed her how to build an audience segment. She asked ridiculously smart questions and we got into showing her how complex it can be, but that at the end of the day we’ve got it under control.
Unlike a relationship where a marketer might say, “I want the agency to handle it and don’t make anything my problem,” we worked together and showed each other where we go really deep into something. At the end of the day we’re responsible for filling in the ad, but we show each other and talk about it together.
We just went into an in-depth discussion about data suppliers and how we want to use them.
BR: It’s our confidence on the ability for Starcom to run the campaigns in a way we expect them to be stewarded. Their ability to understand the details, get visibility at the granular level and execute that vision. It’s been a wonderful partnership in that there’s no black box. We know every detail of the process. We have input into it. We have discussions about it, but we step back and let our agency execute. We have too many brands and too many dollars going into these channels and too many campaigns and too many segments to create for us to do that. We have to let them do what they do best, but we have visibility into the process.
How does that change the nature of the working relationship?
TP: We are allowed to expose problems. We exposed one of our campaigns recently: We tried to test on it, and it didn’t work. Why didn’t it work? I think we set up the Boolean logic wrong. That’s such a fundamental thing: We put the parenthesis in the wrong place in a target we set up.
[Kraft was] like: “Okay.” It’s just that easy. But that’s an intimate detail we were talking about and [Kraft had] an openness and willingness to understand.
BR: It’s also the complexity, understanding things will never be perfect. Every single day, something will surface or something will change. We’ll have to find a workaround or a way to block it. We had one yesterday, which was absolutely new to us in this space. It became visible and we’re addressing it.
TP: You said our KPI was speed. It’s the speed we can address [these problems].
Bob, what do you have against the black box model?
BR: You have no idea what the margins are. You have no idea how they’re defining segments. There are vendors in this space who can’t possibly understand how to define a segment as well as we’d like it defined. There’s no way they can understand who we’re trying to target, what [our customers are] trying to do, what are their motivations or characteristics, the actions they’re displaying that would put them in the right segmentation?
Brand managers know it. The agencies who work with those brands every day and sit in those meetings understand it. But many of the vendors in this space can never identify the segments, the cues and signals that would put them in the segmentation in a way we understand it. And there’s a question of measurement: How is it being measured and tied back?
You’ve mentioned you get your first-party data from sites you own like KraftRecipes.com, which has hundreds of millions of uniques and around 22,000 tags. Can you elaborate on how that helps you segment?
TP: Think of a recipe: It might have cheese or chicken in it. Each of those is one attribute. One attribute: chicken. One attribute: easy. One attribute: takes 20 minutes. So when we run a campaign and look for moms who want easy meal solutions, we look at that data and see what has easy recipes. Philadelphia Cream Cheese launches flavors, and we understand whether a recipe is savory or sweet based on the ingredients in that recipe, so we can use that targeting to notice whether a certain person has a high propensity cooking with sweet or savory ingredients.
How about shopper data?
BR: The industry is changing and the access to that information is increasingly becoming available and to look at this industry and not be able to decision based on purchase behavior is going to become very quickly a detriment to any organization.