Commentary
Why Ad Exchanges Miss The Boat
- by Cory Treffiletti , Featured Contributor, July 29, 2009
The ad exchange model really started back in 1996 with Flycast -- but Flycast was far too early. It promised a self-service platform for agencies to manage their own network buys and people took note. I tested it back then and it was a great solution, but it never saw the widespread adoption that one would have hoped for. This second act of the exchange model promised the same level of access and control on a broader scale -- but the simple fact was that exchanges are not that simple. They're complex, they require training and just as much, if not more, work than the current model for online buy management. As a result, the exchanges have not taken the foothold that they'd hoped.
advertisement
advertisement
There is so much work in the beginning to understand and fully utilize all of the intricate details of the ad exchanges that most agency people have pushed them aside. They require the creation of relationships in the system, management of the relationships and legwork to make sure they are effective.
Exchanges should be as easy as AdWords and AdSense, but unfortunately they aren't. In many cases we've seen ad exchanges hire salespeople to pitch the product and manage buys for agencies, which means they're no different than a traditional ad network.
I was told a long time ago that an idea which saves time is worth money, and in theory the ad exchanges do that, but the promise of the experience didn't match the actual experience, and therefore we aren't seeing the value.
One reason that ad exchanges missed the boat is because they're not filled with premium inventory. I know this statement will generate some angry responses and a defensive stance from people reading this who work at an exchange, but the simple fact is that premium inventory does not make its way to an auction-based model. The inventory is class 2 or class 3, and in many cases is being run from one network or exchange to another, getting passed around more than the donation basket at church. Too much of the time, when you buy the inventory on an exchange you lose control of where it may have originally come from; most advertisers, and especially brand advertisers, are uncomfortable with that level of intangibility and stay away from advertising on these platforms.
What's interesting is that some agencies have been using the exchanges to power their own ad networks, but the development, launch and management of these programs takes time, requires lots of legwork and may not provide a strong enough ROI to be valuable to them in the long run. Just because it's built by the agency doesn't mean the media buyers will be buying it, and if they are pushed or prodded to be buying inventory on those platforms, then that becomes a conflict of interest, and potentially affecting buyers' ability to be objective in evaluating overall media opportunities.
The ad network and exchange space is quite cluttered. Consolidation is, however, definitely taking place as advertisers are focusing their efforts on key partnerships, but rarely do I hear of an exchange as a valued partner. They are interchangeable and expendable in the view of most brands, and not valuable to many others.
Maybe stage 3 of the ad exchange will be more like Google AdWords: a truly self-service, easy-to-use platform where marketers can test out messaging and manage their buys in an effective manner. Either that, or the pendulum is going to shift back to fully integrated, fully customized solutions that require the depth of relationship to manage. I predict the latter. This is still a relationship business -- and no matter what technology comes to the forefront, marketing is about people.
Don't you agree?
Exchanges will never have premium inventory. Premium inventory is best if packaged well - meaning you can usually get a better price than an auction can provide by providing additional value.
As a result, the exchanges will be left to "fill in the gaps", which isn't a terrible role. These gaps are large and need to be filled - so the exchanges play a valuable role. They provide advertisers with relatively low cost inventory and impressions, and provide publishers with steady streams of revenue.
Consolidation and an improvement in ease of use would provide large economies for everyone. That takes time, however. I suspect that as time goes by, this will occur and the exchanges will provide increased value for all.
LOL. I look back to the day you sat across from me at i-traffic and said ".. the problem with Flycast is that it has no place in the current media planning and buying paradigm. ROI is fantastic, but I need to get the plan pre-approved by the client." Stunned silence on my part.
You were right then and I think the same is true today. Exchanges have cleverly positioned themselves in pretty good streams, with good water, and decent fishing. But they're not standing squarely in the stream where the ad dollars flow hard and steady. That requires a real understanding of current process, integration into that process, and creation of frictionless buying systems - not one-offs or obstacles. Exchanges (and many adnets) need to make it simple and easy to buy, measure, and then buy some more.
(I tend to disagree with the premium inventory comments but that's a story for a different day. One that involves publishers holding back premium inventory at premium times and charging rate card if not bump rates for targeted ad buyers.)
Great topic. Exchanges, in whatever form they exist, have limitations. Maybe we're expecting too much from exchanges. Placement control and precise delivery timeline control can be problematic. But if those two dealpoints don't matter, and buyers are willing to manage the bidding process, there are 30+ exchanges out there, not to mention hundreds of ad networks. Caveat emptor, have at it.
Not trying to get too semantic, but maybe the next chapter for us is to move past the discussion of exchanges, and formally acknowledge the term "Marketplace" as the optimal structure for secondary premium avails to get matched to the right buyers. That's where premium avails can and should live (shameless plug: that's *exactly* why we launched my company, TRAFFIQ.com).
A Marketplace is probably the right solution for you if:
--you need to control the delivery timeline (what good is a back-to-school ad that runs in October?)
--you care about the context or environment in which your ads appear, even if you're targeting a specific audience
--you're tired of opaque reports that don't show you impression delivery across all the sites on your buy
Exchanges are a great solution for the right buyer who can thrive in the long tail, or for publishers sitting on remnant avails. But if you're "mid-tail or higher", and these other forms of control matter to you as a buyer or seller of qualified, targeted avails, then maybe it's time for you to get into the Marketplace game.
One could argue that brands are buying on exchanges already - indirectly and perhaps without awareness. The way that brands are buying on exchanges is through the behavioral or social net targeting they purchase for direct response/lead generation. This is complicated buying where the brand is targeting statistically probable people somewhere on the exchange. For the part of the media plan that is for brand awareness and brand engagement, brands care deeply about context and memorable creative - something that only fully transparent premium ad networks or destination sites can deliver. Exchanges have unfiltered and virtually unlimited content. Interestingly, some premium networks (like ones using Adify - but there are others) offer precision targeting on contextually relevant sites - not unlike the advanced targeting in search.
Self-service works in search because it's virtually free to develop as many messages and target key words as you want to test. Then you can target your search display to contextually relevant sites. Display advertising is much more costly to develop unlimited iterations of creative and message to test (with the exception of dynamic ads that just don't seem popular yet with agencies). I suspect that's part of the problem with display self-service for larger advertisers.
Flycast morphed, as you know, from a self-service platform to a premium ad network - the very first of its kind - offering aggregated reach to advertisers.
Although your piece focused on ad exchanges for online media, there have also been a number of failed attempts, some fairly spectacular for online ad exchanges developed for offline media (such as TV and radio). Examples of this would include Google Audio, eBay Media Marketplace, Softwave Media Exchange and years ago AdAuction.com and eMadison.com. I would like to comment on this topic because I have developed what appears to be one of the only, if not the only, successful ad exchange for offline media, Bid4Spots.com. Ours is a weekly reverse auction, where media compete and those providing the lowest cost per thousand viewers (TV) or listeners (radio) win the advertisers’ money. Our growth proves our form of the ad exchange actually works. In 2008, we grew 100% over 2007. By last week, we had grown 100% over 2008 and we’re only half way through the year. How did our ad exchange succeed when, as you point out, so many others failed? I believe three factors are involved:
1. Willing buyers and willing sellers – This sounds obvious but if you examine the failures of online ad exchanges, it usually boils down to the fact that while one side was all for it, the other side refused to “play”. In my opinion, in the case of offline media, the main reason that media (sellers) have refused to play ball in these ad exchanges is that they believe their product will be too commoditized. Offline media sales for radio and television has historically been not only about ratings but also about unquantifiable elements such as listener/viewer attendance at events, community involvement, viewer/listener participation, reputation in the community, etc. When a media property is reduced to its ratings only, the media will feel that it’s an inadequate representation of their property. Online ad exchanges largely reduce media properties to ratings only and are, therefore, avoided by media sellers.
2. Understanding Supply and Demand Dynamics – Forward auctions (where buyers compete and the highest price wins) that drive rates up are anathema to advertisers unless there is a solid reason to pay those high prices. Search engines do this successfully because each PPC advertiser is competing for the attention of a single person searching for that word or term at that moment. Supply is low and demand is high...this is perfect for a forward auction. Lots of demand (from the advertisers) and little supply (only one person searching at a time). So, the advertiser who wants the attention the most will pay the highest price. But in offline media, the opposite is taking place. There are significant amounts of unsold inventory in offline media (high supply) and at the “last minute” advertisers are busy doing other things and don’t have the time or interest to scavenge for deals (low demand). It is for this reason that I came up with the idea of doing a reverse auction (where sellers compete and the lowest price wins) for TV and radio using CPM as the basis for competition for which I have patents pending worldwide.
3. Value Delivered - Walmart has proved it year after year: everyone loves value. This includes advertisers. They want the most audience for the least amount of dollars spent. Driving rates down (through reverse auctions) on top-rated TV or Radio ad inventory that is perfectly good but in low demand yields extraordinary value.
Therefore, I believe Bid4Spots has succeeded as an online ad exchange for offline media because:
• Our auctions are restricted only to “last minute” buys. In this way, media do not see us as a threat; we stay away from scheduled media farther out than “next week”. The fact that we wait to hold our auctions until Thursday mornings for the next broadcast week is testimony to our focus on waiting until the “last minute”.
• Our reverse auction reflects the realities of supply and demand. TV and Radio are just about as popular today as they have been for many years but ad dollars have migrated to more targeted media such as search. This has created an imbalance of where media properties are doing their job of delivering audience but the dollars just aren’t there. We are helping to correct that imbalance.
• Because the auctions are live and only last four hours, there is intense pressure amongst the thousands of radio and cable TV systems to win advertisers dollars. Because our system is CPM-based, advertisers can be sure that they are reaching the largest possible audiences for the least amount of money.
Today's exchanges are crap. Crap stinks. You can put lipstick and a nice dress on crap, but it's still just crap.
Conceptually, the exchange idea should work - but no one has found the magic bullet just yet.
I'm on vacation, laying on a beach working on my already awesome Irish tan and had time to think and write. Let's dig in on some of the larger issues I see with this piece. My general comment is that the exchange space is a Tale of Two Cities - the best of times and the worst of times.
There is no question the exchange concept has taken hold and is changing the way companies operate. It's not happening fast enough though and people come to the marketplace with different biases around supply, demand, data, ad server technology, pixels, pricing, privacy, terms and the future. All have to be reconciled eventually...
No question that exchanges should be as easy as adwords, but the comment misses the complexity of the problem and the evolution of the space. Failed attempts at exchanges first failed because they failed to aggregate enough supply to create the marketplace and in such a way to meet the goals of the buyer. The essence of aggregating supply effectively is putting all the supply and demand in one cookie space with optionality to bid (without all the whacky redirecting Cory mentioned). DCLK failed b/c DFP and DFA were different code bases (hence internal redirects) and because there was no optionality to bid. Fast forward six years to 2007...with Right Media and AdX, we finally see supply aggregated in such a quantity that it makes it a legitimate opportunity, but even those exchanges aren't integrated, so you see aggregation companies popping up saying they will aggregate over the largest exchanges.
After consolidating inventory into one cookie space, there are a number of capabilities that have to be delivered - creative, data/decisioning, supply are all unique capabilities that have to be integrated into one stack and it ain't easy. But there are companies in the last two years, including the holding co agencies that are attacking this challenge. Remember that SEMs built bid optimizing and campaign management tools b/c Adwords wasn't sufficient. So the evolution is happening but not fast enough.
Regarding premium inventory - how do you define it? As an example, a seller of inventory may think an ad placement to a user is not valuable/premium, but that seller may not know that the person, after having visited Facebook 250 times, started the process of looking for a car b/c of Cash for Cluckers, and would now be deemed 'premium' by an advertiser who was able to identify that person as such. Because buyers and sellers define it differently - based on placement, frequency, user targeting (age, demo, intent), etc, it's not accurate to say there is no premium inventory in exchanges. Desirability of the ad is first defined by the buyer. If the buyer has information about a user that defines 'premium' value, does the seller selling it through an exchange make it non premium? For too long, buyers have depended on sellers to define value as context/content and the targeting that the seller provides. That's changing, so watch the concept of premium/non gets turned on its head.
All that said, what is more accurate is to say is that there is a lot of unmonetizable inventory with high frequency and intl geos floating around. That may have been the defining characterisitc of networks and exchanges years ago, but remember that eBay started out selling Pez and Beanie Babies and now sells new cars, jewelry, etc. Same thing happening with exchanges.
There's an evolution in the network space as well. While many networks morphed their marketing to be an exchange b/c it's sexier, a true exchange provides the infrastructure to buy/sell ads regarding of the terms, pricing, placement, transparency, quality. Many networks are morphing to offer their technologies and services beyond their media network b/c a seat at the table as the technology partner is far more valuable than the transient network media relationship. Right Media was the first to do it at real scale in 2004 but many others continue to innovate....audience science, mediamath, traffiq (see their recent PR), turn, collective, specific are all morphing to provide media, technology and services.
And the holding company agencies are not sitting back. They are using exchanges and/or underlying ad technologies to gain unfettered, unintermediated access to inventory with the controls and capabilities in place to more effectively manage ROI and brand, minimize operational expenses, automate the process and build some defensible advantages.
I think your stage 3 isn't actually an either/or proposition. I think it's both, and again, the holding co agencies are leading the way along with some large publishers. There will always be the 'big idea': the one that requires integration, custom development, significant investment, guaranteed delivery, etc. P&G, Mediavest and Yahoo coming together for the award winning program The Thread is a great example. The other end of the spectrum is the automated, audience based, insight driven, bid based buying engine which, at its core, targets appropriate audience in or out of context, controls frequency and pricing and perpetually learns. These two buy scenarios optimally work together.
There are some analogies to buying TV emerging. In TV, a brand may buy 25% sponsorship and 75% audience. The sponsorship is integrated with privileged access, guarantees, integration into content, etc. The audience buys target the audiences using the 3rd party data that defines audiences (i.e. nielsen). In online, we'll likely see something similar where a brand focuses integration efforts with key parnter brands, then uses its own insights or the insights of an intermediary (agency / network) in addition to sellers' targeting to find their audiences.
back to vacation...sorry for any typos, am not re-reading what i wrote.