DSPs Still Fall Short for Brands

Steve Kerho is a 20-year veteran of  the interactive marketing industry and serves as vp of analytics, media and marketing optimization for Organic. Kerho spoke with Digiday about why demand-side platforms haven’t moved much beyond direct response, the lack of high quality content available on exchanges, and the problems with fixing attribution models.

Have DSPs lived up to their promise?
It depends on who’s asking the question. If we look at a particular campaign or a marketer who is using online as a very direct response-focused channel, focusing on the low-funnel, “I’m trying to get in front of people who are in the market for my product or service right now,” I think DSPs have done a pretty good job at delivering. It’s great as an advertiser to be able to sit back and have all of those content providers compete against each other. One of the great things about DSPs is that we are moving closer to my being able to buy a specific audience and I can name my price. I can say this “this is what I’d like the CPM to be” and they all have to compete against each other for that. When I have that kind of transparency I can really manage our clients’ dollars around whatever that marginal return is that we are looking for. And it’s not just about a positive ROI, I can say “I want a $5 ROI or a $7.50 ROI” or whatever target the client comes back to us with. As long as we are working with the lower-funnel they work really well. That being said it doesn’t work as well on the brand side when I’m trying to work higher up the funnel, when I’m  trying to do more upper funnel interactions, creating more awareness, more familiarity. I don’t think it has really delivered what it has promised in that particular state. And honestly we see clients struggle with a fair amount of that. One of the problems about trying to do leading brand advertising is that you want really premium inventory. And there’s not that much premium inventory on demand-side platforms yet. A lot of times what goes along with those brand campaigns is that there is probably a broadcast campaign going on and a print campaign going on and they want the campaigns to all have the same look and feel. And the things that we’ve learned from LOA advertising is that context really matters. If I’m going to be on ESPN, I should really look like I belong on ESPN, which is different than, say, having an ad on something like Daily Candy. So on the brand side you are giving up a lot of that control with DSPs, and so when you are trying to integrate the look and feel of other campaigns, and you know that context matters,  you really want that premium inventory. I think that either way we look at it as advertisers it puts a lot of onus back on us. We have to have a lot of solid performance metrics on the back end to know that we are really getting what we thought we were getting. You can get really, really specific on these DSPs. You can say “my target audience are soccer moms living in the sunbelt who have two kids and live at every odd-numbered address.” DSPs sort of promise you to have that sort of targeting capability. The challenge is “how to I know if I’m really getting that.” While we like the CPMs that come out of DSPs we need really, really good back-end measurement systems. There are a lot of metrics that you have to pull into play to make that happen. You have to look at the search data, the display data, the data that comes from brand.com. We are the ones that end up having to integrate this because it doesn’t happen automatically on the DSP side. That’s the only way that we can figure out if all of these pieces are working together and find out if I’m getting the return that I thought that I would.
Will real brand dollars flow into programmatic buying with the current attribution system?
I think that is one of the major impediments to brand dollars flowing into it. I think the major problem is still the quality of the content. That is first and foremost. The second problem is measurement- can I really measure this in comparison to anything else? Can I look at this in terms of a shared-attribution basis with all the other touch points. Attributiion has been the call of the day for the last few years. Everybody is sort of running around saying, “We’ve got to fix attribution.”  But what we’ve found is that the devil is in the details. It’s not easy to do, it’s important, and it has to be done, but when you do it in direct-response model, it is not nearly as difficult as trying to create a workable attribution model in the brand world. In the brand world I may be trying to create awareness, familiarity, or change and opinion. If I’m trying to figure out how I’m going to share the credit that an ad gets across all of these touchpoints I have to figure out what kind of online behavior I am valuing as a result of that. And in the direct response world that is pretty straight forward; you came to the site, you submitted your email address, you looked at a product,  you bought something. On the branded side, it may be that you came to the site and you may do one of those four things but we have to correlate that with brand-level objectives. We do that everyday, but it is a much higher bar to cross. Those things combined – how do I measure, how do I attribute, how do I find premium inventory? — are the biggest challenges we are facing on the brand side on DSPs.
Can we get closer to solving upper-funnel attribution anytime soon?
I think there definitely is a way to solve this. We have to solve this as marketers because this is where our targets are spending their time. If we could go back five years, there were questions like “can you really do branding online?” . One of the things that helps to work your way through this is a really good testing plan. There are all of these behaviors that I can measure, all of this crazy math, all of this econometric modeling that I can do  to show how online is driving offline behavior. There are definitely ways to get at an understanding of the space. One of the more simple ways to do this is to just take 5 percent of your impressions and just run creative for a charity, say the Red Cross. So now I’ve got a very specific audience that was not exposed, one that was exposed to my ad. Then I get to follow these other behaviors around from my sample and try to figure out it they did anything differently than the other people that weren’t exposed to my ad. Anybody could do that kind of a test. There are a lot of simple ways that marketers can begin to wade into this territory and start to understand what kind of value they are getting, what is this driving for them. You ask these questions apart from looking at whether you want to use a private exchange or be a part of a DSP. These are really just nuanced ways of looking at how much you want to pay for something. It’s really just saying let’s try to understand what these incremental communications are doing for you and how much they are worth.
Do you see private exchanges as a fad or are they a gamechanger?
I think they are really here to stay. I think for a lot of publishers the concern about ad exchanges is that these real-time bidding platforms and exchanges give advantage to the agencies, that the agencies are doing this to bid down the prices of the inventory. Even though the amount of dollars going through these exchanges is relatively small compared to the approximately $26 billion out there in the market, we see it growing pretty rapidly. I think that if they can give us some advantage on the pricing, and fulfill that promise of having premium inventory, I think the exchanges will have a role in the marketplace. Is it going to garner the majority of spend anytime soon? No. I mean people look at DSPs and they see that they promised a lot but didn’t quite give us the whole thing, so I kind of feel like the private exchanges are a reaction to the problems that weren’t solved with the last round of technology that we’ve had. I think that NBCU has probably done one of the best jobs on this, which they did through Admeld. They have people behind the machines that verify that verify that the process is working the way they want, in terms of handling their premium content.
What does the future landscape look like?
On the upside we are going to continue to see spending grow on the digital side. We will still see dollars flow from offline to online. I think unfortunately we won’t see this complexity going away soon. There will be just so many ways to execute your buy, from ad exchanges to DSPs, etc, etc. But we are trying to find an equilibrium. It’s complicated, but there is a lot at stake.