In Search of Exchange 3.0

What's Still Missing to Bring the Biggest Advertising Budgets Online

By Published on .

Andy Atherton
Andy Atherton
The recent launch of DoubleClick's Ad Exchange 2.0 (AdX 2.0) has ignited tremendous excitement over the near-term financial upside in online display advertising. After years of search monopolizing the market's attention, we see the biggest players -- Google and Yahoo, the big agency holding companies and others -- redoubling efforts in display advertising optimization.

The business media is framing the upcoming battle as Yahoo vs. Google, but other players will fight hard too. Why? Because display is a huge ad market and, unlike in search, no winner has yet been crowned. The next 12 to 36 months will determine who takes this next, highly lucrative online advertising prize. However, the outcome won't be determined merely by the short-term competition between AdX 2.0 and Yahoo's Right Media or more generally by any competition between players that only address the direct-response (DR) market.

The DR players (including Google) have been fabulously successful; 30% of all DR media budgets -- some $20 billion -- have come online in only 10 years. But the big-money outcome will be determined by which players can provide the right set of capabilities to facilitate large-scale migration of offline branding budgets online. These capabilities simply aren't present in either exchange or any of the other most talked about offerings today.

Some historical perspective might be helpful in understanding how we have arrived at this point. In the first phase of growth ("Network 1.0"), display ad networks focused almost exclusively on targeting/optimization for DR advertisers measuring performance on a cost-per-action basis. Each operated in a silo, buying and selling as an independent entity. As the number of networks grew, this became inefficient. First, there were many sales and publisher business-development teams replicating the same process of stitching together supply and demand, sending us back toward the 1:1 model the network was designed to solve and resulting in large-scale duplication of DR-focused infrastructure. Second, optimization was siloed also. Only a handful of the largest advertisers and publishers had tools capable of doing even rough optimization between networks. Outcomes for all parties were better than the 1:1 model that came before, but still highly suboptimal.

The "Exchange 1.0" concept, pioneered by Right Media, offered a partial solution to both Network 1.0 issues. First, to the extent the quality publishers participated (which was minimal), Right Media addressed the issue of duplicative negotiations between publishers and ad networks. Networks' sales forces were still duplicating effort, of course, calling on the same clients with increasingly overlapping value propositions. Second, Right Media addressed the challenge of optimizing between networks for participating publishers, but only in a suboptimal way. Networks had to squeeze their proprietary targeting capabilities into Right Media's data model, which meant decreased targeting performance for networks and lower monetization for publishers.

Since Right Media, we've seen evolution within the Exchange 1.0 paradigm. First, supply-side optimization services like Rubicon and Pubmatic are beginning to compete with exchanges. Second, demand-side networks within major agency holding companies are beginning to compete with independent networks. While interesting, these initiatives don't change the Exchange 1.0 paradigm or its focus on DR metrics and functionality.

AdX 2.0 has addressed some of the shortcomings of the Exchange 1.0 ecosystem. The AdX 2.0 real-time bidding (RTB) architecture allows all the cowboys to bring their best guns to the gun fight. Every bidder that can meet the latency performance criteria can submit a bid using all his data on each impression. Bidders no longer need to shoehorn into Right Media's limited data model. Better optimization should also drive higher yields, which may allow AdX 2.0 to attract more high-quality supply. Early indications are positive, but there remains much work to be done.

So what will the future bring? We can envision an Exchange "Version N," where access to inventory is a commodity. Every impression from every publisher is available on an exchange and every buyer and seller can leverage all their data to optimize every ad-serving opportunity. Demand-side networks compete with independent networks to see which can pay the least and still get access to inventory while driving the most direct ROI for their clients. Billions of event-tracking pixels fire every second, microtargeting gives way to nanotargeting and conversion-attribution models become more complex than those highly toxic mortgage-backed assets. Exchanges balance demand-side pricing pressure for publishers by automatically allocating each impression to the highest bidder. All the demand and all the supply is at long last in one big, highly optimized network. Nirvana!

But wait. What about the two-thirds of measured media budgets that are spent against the upper funnel, primarily to drive the nearly 90% of all retail purchases that take place offline? How will we get that offline data into the exchange to take advantage of all of this automated impression-level optimization? How will brand buyers control the environment in which ads are shown? How will they budget and plan in an auction environment without any guarantees? How will they efficiently manage delivery against objectives that can't be measured with a pixel? How will we deal with branding time lags?

The answer is that we and they won't. For example, leading brands are increasingly conducting statistically valid measurement of the impact of online media on offline sales. These econometric capabilities are a valuable bridge until online budgets get large enough to show up routinely in marketing mix models, but they are not trivial to implement. They require a well-structured and carefully managed campaign, with large media budgets for all but the largest brands. The results are aggregated for an entire campaign and delivered three months after the media stops running. Even the descendants of these tools won't support automated creative/placement-level CPA optimization. With the sample sizes, time lags and noise levels involved, they simply can't. And without a tie to offline sales, along with the operational capabilities to manage a top-of-the-funnel campaign, exchange Version N is really just DR network version N+1: The tech-driven Silicon Valley ecosystem has ignored Madison Avenue once again and forgotten about branding. Woops.

AdX 2.0 is valuable in facilitating RTB and getting more supply onto one platform, but brand marketers need different capabilities than exchanges offer today. In order for brands to leverage the emerging exchange ecosystem, they will need sophisticated technology for page-level quality filtering; pricing and delivery prediction; reach and frequency management; composition management; delivery smoothing; and offline impact measurement.

In case it's not obvious yet, that's a very different toolset than the growing clutter of fine targeting and CPA-driven optimization engines. So today's re-energized battle for display is just warming up. The long-term winner will be the one that takes us to the "3.0" ecosystem by providing brand-focused capabilities on top of the evolving supply platforms and helps brand budgets follow audiences online.

As an industry, we can't afford to get this wrong. The 3.0 ecosystem will deliver benefits to all players. Ad Age 100 brands will find more efficient access to their audience in quality content, with a scalable process. Publishers will find better monetization and better creative. Users will find better content, supported by the better monetization that the 3.0 ecosystem provides. And as with Google in search, the players that get it right the fastest will reap outsize rewards.

Andy Atherton is cofounder and chief operating officer of
In this article:
Most Popular