Ad Age: The idea behind ad networks is they make buying easier and more efficient. Are ad networks making your jobs easier?
STEVE USTARIS: Networks make our lives easier for clients that are very, very [direct-response] heavy. If you look at the top advertisers on a monthly basis in digital, it's not really your heavy brand advertisers -- it's the DR ones. And that's why the network space is still hot right now. I'm talking about billions of impressions served over the course of seven days, and you have to hit that Long Tail of the internet to efficiently spend that budget. Plus you need real-time data that you can collect, analyze and act on. You can't do that manually, and the networks, in theory, help you do that.
AUGUSTINE FOU: It's also a great way for us to do testing. I run the test across many ad networks and you see which ones perform better and make decisions based on the mix of sites they have. ... Unfortunately, I don't always know [the sites on which ads will run]. ... They might say here is the list of sites and you tell us which sites you don't want to run on, and then you can assume that the others are the ones you will run on. But that's not actually the same thing as if they showed where the ad did run.
Ad Age:Sounds like transparency -- or lack thereof -- is a main concern.
CHRIS KILKES: That is one of my main concerns. Previously, I've tried to avoid networks as much as possible because of the bad rap they've always gotten. But if I'm going to play with them, I've been very adamant about these specific sites [I want to run on] and that I can also exclude [others].
MR. USTARIS: The push-back we are getting on the client side is true transparency. You disassociate the advertiser from the content, and you start to become a little concerned about the effect that can have on your brand. We do want to be cognizant of content that could hurt your brand ... but clients sometimes start nitpicking out sites that we argue with them have a neutral effect.
Ad Age: What do you mean by neutral effect?
MR. USTARIS: For example, one of our financial clients wants to take out a general-entertainment site because they said it has nothing to do with their brand and nothing to do with clients. And I said that's fine, but I would argue that if the network allows us to target an individual on that site who is in the correct mind-set to make a transaction with your bank, then I don't think the content hurts. ... I would love to be on a site that is contextually relevant to my client, but it's just not possible all the time. And so we have to be cognizant and give the consumer some credit.
NATHAN WOODMAN: From a buyers' point of view, transparency is one of the main issues of why not to go with a network. But to the publisher, a network's having no transparency is kind of a blessing. We probably all read how ESPN basically pulled its ads from Specific Media, and it's along those lines that branded publishers don't want the inventory they release to an ad network or exchange to be identifiable by the buyer because they feel their direct-sales staff will be cannibalized, especially if you allow the buyer to cherry-pick that site.
And that's a telltale sign of an inefficient market. The fact that publishers have unsold inventory means that their pricing strategy is not optimized. And if the publishers would make their sites transparent either via network or through something like an ad exchange, I think you'd find the bidders would be willing to pay more for that.
Ad Age: Really?
MR. WOODMAN: Say ESPN decides they don't want to release their inventory to a Specific Media, but instead they release it through an environment that is more open, like an ad exchange, and identify themselves as ESPN. And then they allow buyers to go in and basically purchase the ESPN inventory. If they did that with a 100% of their inventory, then the market would determine the price, and they would probably sell out a 100% of their inventory and make more money.
MR. FOU: That's perfect price discrimination in effect -- leave it to free-market forces. Publishers are reluctant to do that, but they are not fully taking advantage of the networks if they don't do that.
Ad Age: Well, since you guys brought up exchanges, are any of you using them?
MR. WOODMAN: I have used them. Here is a great example about pricing efficiency and one form of targeting that adds a lot of ad-networks sell, which is retargeting. When you do a retargeting buy, you are agreeing, as the advertiser or agency, to put up one of the ad network's pixels on the client site so that the ad network knows a person has been to the client site. They use that information to basically sell a retargeting list. ... They can charge anywhere from $3 to $5 for that retargeting list. When we went and bought on the ad exchange, instead of putting a pixel up that was owned by the ad network, we put a pixel up owned by the advertiser. We used that information to go to the exchange and say we want to bid on anybody that's on this retargeting list. Since we held better information about the user that is valuable to us, we reduced our retargeting [cost]. Now the volume wasn't there because [the exchange] isn't really super liquid, but at least we reduced our average retargeting price by about 50%. ... The way that I look at it is ad networks that only provide value in aggregations will be disintermediated directly by ad exchanges.
Ad Age: So what a network really needs to survive is some specific targeting technology?
MR. WOODMAN: Proprietary data. And not all of them -- it will be like banking. Right now there are hundreds of ad networks and the only thing that they really provide is aggregation. At some point, there will be six huge ad networks that are aggregation services, and those will probably include the MSNs, the Googles, Platform A and Yahoo.
MR. FOU: Yahoo has a lot more personal information through its other services for which you registered. So they can cross-target with demographic information ... and because Google doesn't have similar information, Yahoo actually has better proprietary data at this point in time.
MR. WOODMAN: And should therefore receive higher bids and, over time, higher yield.
Ad Age: You bring up the question of contextual vs. other kinds of targeting. What works better?
MR. WOODMAN: I think it's better framed as a matter of more perfect or less perfect information and which information is more usable for the buyer. Registration, contextual -- they are all a piece of having perfect information, and it's about which one of those pieces of information is more valuable to the buyer.
MR. KILKES: The power of optimization is that you can test all that stuff. We've seen that Yahoo's registration offering leads to much more engaged audiences vs. what we have see through, say, a Google gadget. That leads us to believe that combining registration data with behavioral is just narrowing the funnel a lot more efficiently for us.
The biggest problem with testing different kind of behavioral segments is maybe we are only reaching 10,000 people on a specific site, like NYTimes.com. When you are talking about reach and trying to get frequency, that becomes pretty problematic. Why do I spend 10 grand on 10,000 people? Of course you would say, "Well, that's a richer audience." That's one of the big challenges for behavioral is you're narrowing your universe so much.
Ad Age: You talked about networks as a direct-response vehicle. Are you getting the sense brands that are less DR-focused are starting to use networks more? Can they make their service translate to a brand that's not a DR marketer?
MR. FOU: Let me use an example from the consumer package-goods market. Historically, CPGs have not done a lot online because they are saying you still have to buy the Campbell's soup in a grocery store -- so why would we do that? ... Right now, the data from the point of sale, like what Catalina might have, are hard to marry back to the data we get online. But there are techniques to kind of bridge the gap, something as simple as a click-and-print coupon to get you to the store to buy.
Ad Age:Is there branding value to advertising via a network?
MR. USTARIS: Sure, here's an example. For our client Reebok, we were trying to get people to visit rbkcustom.com, where you customize your sneakers. And through research we found these consumers -- this fashionista sneaker-head -- are extremely fragmented when it comes to online choice. ... But there are specific categories they're interested in, like snowboarding and sneakers, obviously, and music. So we understood we had to reach this individual not on a Yahoo or an MSN but more in the Long Tail of internet, and we had no method of really doing that. That's when we said to the ad network, "OK, here are the categories. Here is what we think demographically this individual looks like, so put that in your system."
And then it was incumbent on us to put together a creative piece that would be curious to these individuals, because they are not really about mass marketing. So our creative team put together this piece that was unbranded and just asked you a question like "Choose a color." You didn't know what the endgame was, and there was no RBK logo on it. The theory was that the network would start to narrow down that list of sites. We go broad with our assumptions, and the network does the job of, based on interaction, optimizing that list.
That Long Tail gets smaller and we are focusing on one who the sneaker-head is and where they are. And it worked really well for us, not only in getting visits there but keeping the post-campaign traffic at a higher level than it was pre-campaign.
Ad Age: Did you take any of those insights that you got from the network and apply those in other media?
MR. USTARIS: We didn't, but if I could go back, I would probably have approached a Tacoda or Revenue Science and asked them if they could do that post-analysis and ask them to help us profile these behaviors a little more so we could understand them better.
MR. WOODMAN: And that post-analysis that an ad network can do is largely dependent on how their optimization operates. For most of them it's a black box. ... Not really many people even at the ad network know what it's doing because nobody ever sees it. But then you start layering other types of data like context or behavioral profiling, and you can do the post-campaign analysis like what Steve is referring to. You get more transparency in terms of the audience you are reaching and engaging.
Ad Age: What about all these publishers launching vertical ad networks -- for example Martha Stewart launching Martha's Circle? Do you see a need for that?
MR. WOODMAN: In categories where you know you may be getting people who are close to purchasin -- say the Kelly Blue Books of the world. ... There are certain verticals that have heavy demand. And so in those instances, for those types of companies to kind of extend their reach by these virtual networks, then it makes sense.
I don't think that some of these Martha Stewart plays are all about necessarily expanding inventory. I think that a lot of that has to do with expanding their data network and increasing their reach and starting to do some of their own type of behavioral profiling. ... That's a secondary and often unsaid benefit that might eventually outweigh the primary benefit.
Ad Age: Now that the Google-DoubleClick purchase finally closed, how do you see that affecting the AdSense network?
MR. WOODMAN: I have a theory here, as an outside observer. Google has always had an issue showing display because four very-well-placed text ads take up the same space as one display ad. And the probability that somebody will click on one of those ads vs. click on the display ad has pretty much always won, which means Google's yield and the publisher's yield are higher when they serve text ads.
So it's really just how the incentive structure works within Google that has made it difficult for them to have any display inventor -- unless they clearly just market it as display inventory and don't allow the text ads to compete against display.
I think Google's text bid will compete against DoubleClick's display bid in their internal auction system. DoubleClick will be unchained a bit because Google has never been able to do cookies or behavioral profiling or demographic data that Google doesn't want to get involved with.
Ad Age: Does it bother you that DoubleClick and Atlas are both owned by media sellers? Do you want them to be the ones giving you all the ad-serving data?
MR. KILKES: It definitely bothers me a lot, but in terms of what options are out there ... they're really small.
MR. WOODMAN: It's really interesting to talk to the platforms and try to get an indication about what they say but also what they do. You start getting a sense of whether or not they intend to be open or intend to create a proprietary system. ... As long as platform players talk about wanting to remain open and transparent, give complete access to all buyers and sellers in the system and create a market-based transaction system, I am not yet concerned.
Ad Age: Steve, Carat has its own ad-serving system, right? Was this in response to the sellers snapping up these existing systems?
MR. USTARIS: Aegis just recently purchased Blue Streak ad-serving system. It sort of was a response. You just never know what limitations they're going to put on you, and so we looked down to see what else was available, and we made the purchase.
Ad Age: If you could outline your 2008 wish list for the ad networks, what does it include?
MR. WOODMAN: I want to drive home the issue of transparency -- and transparency in a number of different ways: pricing, the site that you are running on. I think networks will eventually need to have a completely open, transparent, market-based system, which will provide higher yield for publishers, for intermediaries or anybody.
MR. KILKES:: My major concern is the privacy stuff that's bubbling up, and it's not just for ad network. It's for online as a whole to address responsibly, because I'm concerned this cookie stuff we are doing for ad networks can get shut down pretty quickly if we don't address that in a smart kind of open manner.