Anthony Rhind is co-chief executive officer at Havas Digital and, given his senior position at one of the biggest agency holding companies, has an excellent perspective on how big agencies are developing strategies for exchange trading. Rhind took time to speak to ExchangeWire this week about the challenges of buying media through automated platforms, the complexities of managing client and agency data, and what impact the DSPs are likey to have on European market.

Do you think the large European holding companies like Havas are now seeing ad exchanges as an efficient channel to buy ad inventory?

AR: The ad exchange space is still nascent in Europe. Nonetheless we believe that flexible inventory marketplaces supported by combined open and proprietary data query, ideally in real-time, will improve the efficiency of both buying and selling of ad inventory. We think the opportunity for Havas Digital clients will achieve scale in 2010; initially the focus will be on buying efficiency but our goal is to develop the expertise and platform infrastructure to scale behind performance effectiveness. This will ensure publishers are incentivized to support growth and volume rather than fear negative pricing pressure.

Do I think this vision is shared across the large holding companies? No I do not.

Do you think that trading on ad exchanges makes it easier to leverage agency and client data to deliver better campaign performance?

AR: Ad exchanges make it possible to leverage agency and client data, it is certainly not easy. The first challenge is to assemble the data in a manner that allows it to be matched to the available inventory to direct valuation, bid execution and then instruct creative. Bear in mind, disaggregated data is required to match data sets and impression opportunities on an individual browser / cookie or other unique identifier basis. This involves massive volumes of data and significant system complexity; we are the only agency with a history of managing this in-house using our proprietary Artemis data platform.

The second challenge is to manage data complexity. Once you have that data, in itself no small task, you need to know how to interpret the relative contribution of each variable and understand the opportunity cost for each impression investment. Consider the current industry attribution remains ‘last event’, ignoring frequency or format, let alone environment; I am not confident many in the industry are ready to take advantage of the data nirvana which is so exciting the industry. We have demonstrated a willingness to address the complexity of attribution process and we are testing sophisticated trading algorithms. I believe the data informed ‘decisioning’ process rather than simply data access will be the key to campaign performance.

The third challenge is to connect the data and the decision to the inventory marketplaces using a platform which can operate quickly enough to query the volume of impressions available within the ad-call window, at most 150 mille-seconds. Given the volume of impressions available, currently around180,000 impressions per second on RightMedia and 30,000 impressions per second on AdX, it requires both incredible performance and massive scale.

What in your view are the key advantages of buying inventory through the ad exchanges?

AR: The obvious advantage is the ability to segment inventory by impression, simplistically the idea of buying ‘audience’ not ‘placement’. In fact it can be either or both, in combination. It becomes possible to execute retargeting with both scale and speed, recency can be managed when you proactively seek to re-message to an individual rather than waiting for them to return to a known environment. Variable pricing is an advantage, at this point mostly weighted to the buyer. However as publishers learn to enforce scarcity (rather than wallpaper sites with advertising ‘opportunities’) it will help publishers manage yield upwards. It is actually because I genuinely believe segmented buying and variable pricing will lead to increased yields for publishers that it is essential that the ad exchanges / inventory access platforms have transparent pricing and price reduction auction models.

Flexible segmentation, variable pricing and media / data unbundling bring certainly to advertisers, this will stimulate advertising yield for publishers because advertiser ROI will improve. The existing model of ‘cheap’ should finally be replaced by effective.

Brand safety is a primary concern for most European advertisers when running display campaigns. Is enough being done to ensure brand safety on ad exchanges?

AR: This is a critical issue for our industry. It applies to ad network buys as much as ad exchange buys, in fact it can be an issue with run-of-network portal buys. Given the size of campaigns, we tracked in excess of 150 billion impressions in 2009 for our clients, it is impossible to guarantee everything is served as bought without a technology layer. We work with a number of specialist providers across our markets in this space, one AdSafe is also talking to IASH about an ‘industry level’ initiative. However, as is often the case, it comes down to whether the advertiser would prefer to buy ‘insurance’ or put all budget into buying media without attendant protection.

Is placement and context less important than audience? Can this be a key driver in shifting spend to ad exchanges?

AR: That will vary by client, campaign, individual consumer and very explicitly it will vary by point in purchase funnel at both the product level and the individual consumer level.
Ideally advertising should be targeted to the correct person, at a moment when they are prepared to give attention, ideally in an environment which helps them register the relevance to them … very importantly, with a message which is effective. If this nexus is incomplete, trade-offs are required. As I mentioned earlier, there is so much complexity I believe algorithms should help us manage the opportunity cost for each impression opportunity. Environment / context is critical, we are working on a way to feed this into our data set alongside audience variables to direct our ad exchange buying.

Do you think brand campaigns could work as well as DR on ad exchanges?

AR: Absolutely. In fact ad exchanges need brand advertisers to want to buy segmented audiences in order to make more publishers understand that flexible buying will help them better monetize both their audience and their most highly demanded ‘premium’ placements.

Do you think the arrival of DSPs into the market will help agencies bridge the technology and skills gap?

AR: The US is certainly driving this area for reasons of digital marketplace scale, well established tech capital investment structures and the presence of technology entrepreneurs on both West Coast and East Coast. The plans of a number of the US DSPs to launch in Europe will certainly accelerate the sophistication of the European ad exchange marketplace. However, Europe isn’t the same as the US and I think the much smaller scale of the local markets combined with incredible competition will shock the US pioneers.

Many of the DSPs have evolved into service models from platform strategies, I think the US DSPs will actually ultimately decide not to focus on building a network of ‘DSP agencies’ and instead prefer to scale behind a licensing business model.

As we are a global holding company with a very strong European influence, we have been developing our DSP Adnetik as a global operation from conception. We have managed ad exchange campaigns in many markets in Europe, Asia, LatAm in addition to the US to develop and refine our product and service. You can expect to see a concerted roll out across this year.

At the recent AdMeld RTB Forum event you said: “data is scarcer than media … the price for data may exceed the price for media down the line”. Do agencies realize the value of their data and are we close to the day when data will exceed media in value?

AR: I have touched on this theme already, specifically the readiness of agencies to understand the value of data. First I must say, the data belongs to the client not the agency. Second, I would like to expand on the statement.

At the moment media is not scarce, specifically there is far more supply of digital advertising inventory than there is demand. The oversupply is growing as the internet universe expands through universal access programmes in Western markets and expands through infrastructure investments into emerging markets. The oversupply is growing as the global economic crisis of the last 18 months reduces per capita spend on discretionary items, especially social and entertainment … people are spending more time on fixed cost entertainment such as TV and web.

In parallel marketing adoption of digital still lags consumer adoption, TV has never been cheaper and marketing budgets are certainly not growing universally. So some media has very low value, if data is added to qualify the audience attached to the media (impression) then the data is creating the value. This is already the case, it is a foundation of the ad network business model and less established, the genesis of emerging data trading companies such as BlueKai, eXcelate and others.

What’s your view on RTB (real-time bidding)? Is it a more efficient way to trade media? Will it become a standard way to trade on automated platforms here in Europe?

AR: Our early results from RTB show very strong performance gain versus standard ad exchange buying, albeit at this point all experience of RTB has been within the US. As we bring more data points and more varied criteria into the real-time decision we anticipate further ability to refine segments and define value.

Given the many variables in any campaign, the value of data can vary from one campaign to another, but the emergence of real-time bidding offers additional ways for marketers to apply the data. These include cookie segment targeting, use of first-party data to inform bidding and audience extension that combines both techniques.

Real-time bidding can also significantly enhance the value of data in a few key ways such as enabling greater opportunity to reach important segments; this is because the large audience reach offered by exchanges and other aggregated inventory sources increases the likelihood of reaching desired consumers and managing frequency to effective level. Also, the optimization advantages offered by the impression-level pricing of real-time bidding, can make finding and buying your segment more cost-effective.

We believe it should be a standard that inventory can be accessed on a real-time bidding basis. However, I am not sure how quickly a majority of the buy side will be able to manage data update in real-time. As I mentioned earlier, we are the only agency with a history of managing ad-server data on a disaggregated level. Within the last 12 months one or two other holding companies have talked about developing such capability, you will need to ask them where they are in this build process but it is far from a straightforward development project.

This means that aside from Havas Digital and Adnetik, the majority of the capability on the buy side to execute real-time bidding sits inside the ad networks, the supply side optimizers such as AdMeld plus one or two of the leading US DSP companies. Until more agencies can execute RTB, it will be a major competitive advantage for agencies like us who have moved first to develop RTB capability.