The Official Blog of Cogmap, the Org Chart Wiki


Archive for July, 2009


Can Ad Networks and Premier Inventory Co-exist?

Monday, July 27th, 2009

Jon Miller and Ted Leonsis love the Ferber brothers!

AOL Advertising (formerly Platform-A) is watching the pendulum swing to a sales model that segments owned and operated inventory sales from the network.  After a very brief experiment selling them together, they will now be sold separately again.

Before I address the question of whether it is possible to sell both of these things together, let me give my quick perspective on some of the organizational history of AOL Advertising.  None of this is particularly interesting and I assure you that none of it is revisionist, but this is the opinion I had for more than 2 years as we tore through leader after helpless leader.

Randy and Ron were looking for someone to combine the sales forces of Platform-A. was growing like gangbusters and AOL ad inventory sales were flagging.  AOL sales had all of these senior, senior guys that knew every player in the market, had great experience and relationships, and was missing numbers left and right.  Randy and Ron’s theory,  I tend to imagine, was that the difference must be that the network product was simply better.  People were buying it despite a more junior, less-experienced sales force.  If we had our best people selling our best products, think of the amazing things we could do!

I was not that close to sales or Lynda, so I don’t know for sure, and I never had one conversation with Randy and Ron, so I really, really don’t know, but I figured that when Mike Kelly was getting pressure from Randy and Ron to combine them, was fighting to not combine because there was really nothing in it for them: combination probably would have meant that it would be harder to hit their numbers, their more junior people would be the odd people out in an account re-alignment, etc..  Furthermore, was the golden goose, no one wanted to mess with it, and Mike Kelly, a long, long, long-time media sales guy and a good guy all-around (although I never had one conversation with him, he seemed nice enough and smart), knew, as I did, that sales force re-orgs always mean 2 quarters of missed numbers no matter what.

Mike fails to act and Curt gets the job.  Curt doesn’t do it, so Lynda gets the job.  At this point, Randy and Ron were probably hoping that if Lynda was suddenly responsible for the whole number, suddenly she would be a lot more open-minded about salesforce integration.

Yea, and verily it came to pass, there was one sales force selling products virtually instantly.

Yea, and verily it came to pass, quarters were missed.

Yea, and verily it came to pass, management whipped and sawed and everyone was replaced yet again.

I guess this article implies that I don’t value sales reorganizations.  Let us be clear, I do value them, but only if you really know that you are moving to a much better structure.  It better be a lot better to justify the two quarters you instantly miss on the one.  Furthermore, you rarely, rarely really know if it is that much better, so it is typically hard for me to justify.

Hopefully this also makes it clear that, in my opinion, there was probably not much anyone could have done to yield a better outcome in this situation.  In the abstract, everyone did the best they could with what they had.  Certain players could have tried to do things different, but the end result probably would have been much the same.

Anyway, back to the original question: Can a company even sell premium inventory it owns and operates and run an ad network?

It is hard.  Part of being an awesome ad network sales force is to tell people that they overpay every time they buy inventory directly.  Talk about how much more efficient ad network purchasing practices are.  The power of reach, optimization, and tight frequency caps.

All of these are anathema to selling your own inventory.  And the closer the two sales forces are, the harder it is to really go after each other – and that is exactly what a sales guy has to do to be successful: Killer instinct every day.  Ad network guys have to bash home page buys.  People selling a home page have to deprecate networks.  It is their nature.

Chinese Walls in Media Buying Can’t Make Anyone Feel Good

Monday, July 27th, 2009

Was reading AdExchanger today, like every good digital arbitrageur, and thought he had some interesting highlights from iMedia Connection’s recent article by Tom Hespos.

Tom expresses concern that starting an ad network can create conflicts of interest for agencies.  A Razorfish-ian responds back that they incorporate them separately and that addresses the conflict:

“I don’t think anyone disagrees with your point, ‘you can’t ask a media seller to solve high-level business problems.’ Most of the agencies who are pursuing this opportunity break out the inventory-selling part of their business as a separate company (Havas’ Adnetic, IPG’s Cadreon, Razorfish’s ATOM Systems, etc). I think this resolves any conflict of interest, and cleanly separates the strategic/planning arm of the agency from the transactional arm.”

My immediate reflection on that remark was: “Yep, that is what they said about analysts on Wall Street during the first Internet bubble”.

I do have to say, I think we have already seen some of the challenges in making that work.  Let me illustrate a couple of examples of theoretical problems that maybe have manifested themselves in real life:

  1. Would another agency ever buy from an agencies network?  My impression is “no”.  Without aggregating demand, can an agency actually achieve the scale necessary to have strong network reach, much less the volume of behavioral data necessary to build true behavioral scale?  It seems unlikely.
  2. Clearly, the only hope for this is to force all of your advertisers to buy via your platform.  The key to enabling a lot of volume via a single platform is complete transparency.  If site/placement details are hidden as part of the buy, then buying the network becomes “one of the things you buy” rather than buying many sites via a network.  If there is complete transparency, then the publishers should be requiring market rates, destroying the network leverage.  If it is rolled up, then it becomes a tiny part of the buy.  I believe that all of this is part of what happened with DrivePM, which seemed to me several years ago to have the organization, team, relationships, and opportunity to become a major player.  Unfortunately, they disappeared never to be heard from again.  While I never heard any details, I always felt like being a part of an agency held them back.

The Impracticality of Optimizing for Interests

Monday, July 6th, 2009

Behavioral targeting is a rapidly growing part of the online advertising mix.  Given this, I wondered, would it be practical to build algorithms focused solely on the effectiveness of behavioral data?

In a world of CPA, certainly.  In a world of CPC – the world of smaller advertisers – not so much.  The key problem is a lack of effectiveness data.  Optimizing inventory is easy.  You own the ad space where the ad gets shown, you know if someone clicked or not.  Behaviorally, you have to acquire the behavioral data and then also acquire the ad space to find out if someone clicked.  The inefficiency created in generating learnings seems to me too large.

Sorry I don’t have time for a more thought-provoking version of this post.