Display Advertising’s Killer App

Posted: November 24th, 2009 | Author: bloom | Filed under: Random | Tags: ad exchanges, ad tech, adgear, advertising |

Every once in a while you read something that just sticks with you. At the time of reading you don’t notice it, it’s part of a myriad of other media snacks you consume in a day. But a few days later you realize that this little something has stuck and is now connected to other dots in your head.

Russell Davies’ talk at dConstruct 2009 — and the following post on his blog — was exactly that for me. The talk itself is great and makes you wish you were there. But I digress. Russell talks about a number of things, but what stuck with me was the bits and pieces he got out of a book called The Nature of Technology, by Brian Arthur. The book explores how technology evolves, the mechanisms and physics of how these changes occur. The way these rules relate to the evolution in display advertising is intriguing to say the least.

Display advertising has lost a lot of its luster in the past few years, where search obviously took over and became a larger component of online advertising budgets. It is not a secret that now about half of all internet ad spend goes into search. Some say that it is the future. I tend to disagree and here are two main reasons:

• Brands are more important than ever, and brand building can hardly be done with 3-line text ads that sound like haiku. Advertisers will continue to reach potential consumers in rich visual environments in order to create an emotional impact. It is experiences that create brands. Think of Apple’s paint splashes and multi-colored iPods on billboards. Or a BMW roadster driving through the loopy roads on a mountain with a view of the ocean. Or a Louis Vuitton bag on Sean Connery’s shoulder. You get the point.

• There is great scarcity when it comes to reaching people who expressly manifest their intent on search engines. There are only so many people that will type in “fancy restaurant New York” on Google.  Advertisers will always want more of it. Not only that, but they would want to reach those that their competitors can’t reach. They would want to create a competitive advantage instead of just bidding more than the other guy. This is hard, if not impossible, to do in search. And it would be to Google’s disadvantage, since the price of a click goes up only when all parties have access to the same information about the user. If an advertiser is the only one bidding on a keyword, it is not good for Google. But more on that later.

In the early days of display advertising the industry simply emulated traditional media, buying impressions in bulk under the dark veil of the empty promise of a “better, completely measurable medium”.

Who are we kidding? If you’re as disappointed as I am, you know that this is exactly the way online media is still bought. In the words of Eric Picard (whose latest article in MediaPost is worth reading), agencies are hiring armies of liberal arts majors and arm them with massive budgets and negotiating power. The result is 6-week media plans set in stone. Hundreds of millions of impressions being bought in the Sports section of a site or in the Sports channel on an ad network. CTR being the only metric watched closely and optimized against. Tags being sent via email by the agency. Discrepancy resolution between ad servers. We have no idea where in the person’s day our impression fits in. Did we serve it to someone who searched for tomatoes on Google and somehow ended up on the site we bought? Was the creative even seen on screen, or did it stay below the fold without the user scrolling down? In video advertising we don’t make a difference between a video of a parrot on a skateboard and quality content on, say, blip.tv. Video is treated as a form with complete disregard to content and type of content.

Media planners and traffickers have enough trouble making sure campaigns go live on the right date and there are no glitches with the creative. They don’t have time or technology to answer all those questions. Display advertising, as it was done until now, is broken and you know it.

One of the most powerful concepts that Brian Arthur explores it that of redomaining, or introduction of new components and new ideas from a completely different technological domain.

Search marketing, of course, is itself a good application of redomaining. In its domain where resources (people who search for a certain word) are scarce, it makes sense to use an auction model. Thus, people searching for something specific are kind of like Van Gogh paintings, and Google is kind of like Sotheby’s. It just makes sense for scarce resources.

Data as the Intel Inside

Over the past few years, ad agencies started doing more search internally. Many large agencies have acquired specialized search shops and rolled them into their media departments. Search teams have started hanging alongside display teams. Typically, clashes have risen where search people just don’t get how you can spend millions of dollars on ad impressions without really getting much proof that you’ve reached the right people. They are impressed by how little automation exists, with tags still being sent by email in zip files. And you can’t really roll up your own systems that aggregate data by using APIs provided by publishers and networks.

But sparks have started flying because  some of these search concepts are being redomained into display. More specifically, the emphasis on data and accountability for every “thing” you pay for. This may not seem like that big of a deal, after all we’ve being doing “behavioral targeting” for so long! But this time it’s different.

New technologies are emerging where an agency ad server can talk to the publisher ad server before the impression is served. Not only that, but the agency’s ad server can also tell the publisher ad server how much the impression is worth to this agency.

The pitch to agency’s clients is quite powerful: with this, we only pay for impressions we’re absolutely sure we want, making sure we’re paying exactly the right price. For every single impression, the agency calls the shots on whether it wants to serve to this user, and if so, how much it wants to pay.  No more waste. This is very exciting and it’s what the second generation of ad exchanges is about.

But what criteria can agencies use to call the shots on each impression? Of course, there are some intrinsic variables such as the geographic location of the user, global frequency cap across the whole media buy, time of day — but those are part of the traditional ad serving past.

The real magic, however, happens when an agency has some data on the user that others don’t have. The agency can now witness visits to its advertisers’ sites, and using ad exchanges, retarget them later on, with ease, across millions of sites and billions of impressions.

Isn’t it just good business sense to show a good deal to someone who has visited a telco’s site in the past few days? Another range of criteria for targeting are made available from the trend where data is being decoupled from its original media context.

Companies like BlueKai and Exelate make witnessing agreements with, say, Expedia. If someone looks up flights to Las Vegas on Expedia, BlueKai makes that data available to agencies.  The agency can then target people who have looked up flights to Las Vegas on sites other than the original Expedia. Buying this data and overlaying their own business rules on top, or creating direct witnessing agreements with publishers where intent can be harvested, ultimately creates a solid competitive advantage for the agency.

Data as the Intel Inside, although it may seem to be part of a web 2.0, a bigger metaphor that has lost its meaning, is shaping up for display advertising in a major way.

What does this mean for agencies?

Media planning is quickly becoming a complex science where agencies must go beyond traditional ad serving principles of “put in creative, check comscore, negotiate with publisher, configure placement, send tag, get report”.

To address these challenges, agencies have started developing their own demand platforms, or software platforms that aggregate data and connect to ad exchanges and ad networks to gain access to targetable inventory. They have started buying audiences instead of inventory, and they are the ones making sure that the data is clean.

Although relatively new, demand platforms are in operation at Varick, WPP, IPG, Havas and closer to home in Canada — at Cossette (built on top of our platform, AdGear).

The advantages for agencies are clear:

• The agency has the assurance that the data used for targeting is clean — they don’t have to trust a third party or an ad network (who sells the same thing to the agency’s competitors!).

• The agency creates a range of competitive advantages by developing infrastructure that provides insight and optimizes the ROI in real time.

• The agency gains strategic control over client accounts as data and insight are difficult to migrate. This, of course, also raises the questions on data ownership.

The fact that agencies can bid on every single impression in real time adds an interesting dimension to the problem, since one agency can know something about the user that another doesn’t know. Or it may want a user for reasons other than know to others. This creates what is called an asymmetry of information. Ultimately this is what makes this different from a stock exchange model, where all auction participants know exactly the same information about a barrel of oil.

What does this mean for publishers?

Publisher’s job is helping brands create experiences that make brands stronger. But how can you do it if, as a publisher, you’re no different from another? Betting on deep creative integration, with exclusive ad formats and branded content is a solid track that will only become stronger. Formats should get bigger, better, richer, more overwhelming if they are really to work for brands. But this can only scale if the right data is there to back it up.

The opportunity not to be missed by publishers lies in making data work for them. Exploring data in order to understand the audience the way agencies armed with demand platforms understand it. Imagining what would happen if agencies selected every impression, and making sure that the right premium is paid for such precision. Because there is no way back. The days of undifferentiated media buys are coming to an end.

This means better technology, allowing for data reporting and allocation of inventory in an ad exchange model. This may be for ad exchanges such as the DoubleClick AdEx, but it could also be their own platforms. As Varick’s Darren Herman recently mentioned, there is a trend of publishers developing their own platforms that work like ad exchanges.

When we first started working on AdGear, we imagined a world where ad platforms talk to each other. It wasn’t very clear how it was going to happen, but our gut was telling us that the future of advertising on the internet is not one SkyNet ad network. Instead, ad infrastructure would become more like the internet itself — a loosely coupled organism with synapses spreading in all sorts of directions, sending electric charges back and forth. In our case, the electric charges communicate availability, price and insight.

Now the time finally came when this transition to connected ad architecture is starting to happen. Billions of impressions are already bought and sold this way and a new generation of ad serving platforms allow to buy, share, analyze data and manage services that have access to inventory (like ad exchanges, ad optimizers, ad networks, etc).

This is by far the most exciting time in display advertising’s short history. Publishers have a chance to participate in the creation of a profitable model where experiences created on their properties are backed up by data. And although the industry still faces many challenges, the transformation of undifferentiated pools of inventory into quantifiable audiences is one big step in the right direction.

---