Ad Inquisitor

Leverage: Who has it?

Here's an interesting question: Who has the real leverage in online media?

Hypothetical situation. A large direct-response advertiser decides to implement an optimization solution - let's say Poindexter. It calls up FastClick and says "Here's my new Poindexter tag". FastClick, worried about the new behavior network that Poindexter is launching, says "We don't support Poindexter. We'll only accept raw creative." What happens?

Another hypothetical situation. A major publisher, one with a lot of remnant inventory, purchases Atlas to optimize and manage this inventory pool. Casale Media, threatened by the competition with Drive PM, tells the publisher that they won't serve ads that come from Atlas. What happens?

Both of these situations occurred recently - with major players - and here's what actually happened: The advertiser laughed at the network and said "I spend $500k a month. You'll take whatever tags I give you." The publisher laughed at the network and said "Fine. If you don't want the traffic, I'll send it to somebody else."

I think it's pretty clear what's happening. Not only are networks undifferentiated, they overlap heavily with each other. Smart publishers are implementing their own monetization programs to work directly with direct-response advertisers. Networks have no leverage. What's worse for them, they're under direct attack by Google.

Here's a pretty no-lose prediction for 2006: Networks are going to pay much more for inventory than they ever have before.

Here's another: If networks don't find a better way to cooperate than daisy-chaining, they're toast. Google is big enough and good enough to beat a bickering set of networks - and advertisers will start working directly with publishers' monetization programs.

Leverage is a bitch... if you don't have it.

Why NOT to "optimize"

Towards the end of the boom, I went to a seminar that one of the One-to-One marketing gurus held - Rogers or Peppers or somebody like that. At the Plaza in Manhattan, lots of execs around, and the thing is sponsored by some enterprise software company, maybe BEA, who had helped a couple of companies go one-to-one with great results. The speaker was talking about how important it was to take a fresh look at your business and restructure the whole thing, top to bottom, to be able to deliver customized products. Applause, happiness, buy-in.

Then BEA's speaker started talking about some case studies and success stories.  There were two types: companies that were so desperate that they couldn't help but get better with a relatively simple implementation, and companies that redesigned themselves to embody the one-to-one mentality. From BEA's perspective, the quick fix sounds great. Install our magic product and you'll make millions! And of course the companies that rebuilt their businesses around personalization made BEA look fantastic, even though BEA probably didn't have much to do with it.

The elephant in the room was everybody else... all of the companies that were doing well enough that a quick fix wouldn't be possible and the pain of restructuring didn't sound so attractive. Lots of head nods from this bunch. Not a whole lot of compulsion to act... and they didn't. And now personalization has faded away and we're on to the next set of buzz words and fancy speakers.

So how does this apply to online media? I think that the way the "optimization" vendors have sold themselves is a lot like how personalization vendors were selling themselves a few years ago. It sounds great; it works in a few desperate cases; but really isn't compelling on its own - you have to change your entire philosophy to make it really work.

A great example of this is the DFP optimization product. I'm not privy to all of the details, but I think it's fair to make some assumptions. DART works on a priority basis. When people want to optimize using DART, they take a spreadsheet and figure out which deals work on which inventory, and upload the spreadsheet into DART as priorities and maybe frequency caps. DART's optimization tool does this automatically.

You know how Poindexter's tools work? They take a big old dump of server logs. They run it through SAS. They take the statistical outputs and build some kind of targeting matrix. The ad server chooses ads based on the matrix.

How about Arbiter? Same thing, more or less. How about SDC? Same thing, different day.

So wait, Mr. Inquisitor. These guys have been telling me that these tools will make me rich. Hotbar loves SDC. MSN loves Arbiter. AOL loves Poindexter. (nobody loves DART yet - but somebody will). So what gives? Shouldn't I be talking to these guys?

It's a lot like personalization. If you have a really big problem - like a few billion completely undermonetized impressions - these tools may be able to help you. They're a bit better than the spreadsheets you're using now. But they're not going to help you build an efficient business, and in fact, they're more likely to hurt you in the long run.

Great example - if you want to buy media from MSN, they have to first qualify you into one of their media programs. Either you buy performance media (powered by Arbiter) or search media (Overture/proprietary) or premium media (internal). If you want to buy performance, you get a great PowerPoint deck and a totally incomprehensible "bid guide" - basically a rate card on steroids. Three months later, after you negotiate (negotiate? but I thought it was a bid guide?) and get approved (but I use a standard 3rd party server?) and renegotiate (don't ask) you may be able to get your campaign running - but no guarantees that it works, since they can't manage your campaign to an ROI goal unless you buy CPA.

So a few customers wind their way through this mess and place a buy. And MSN throws a party, because they're making money where before there was nothing. And great rejoicing at Arbiter; they go try to sell the same thing to the other desperate housewives in the industry... and make no headway, because SDC and Poindexter and all the other "optimization" companies are all looking for the same low-hanging fruit.

The problem is that this distracts MSN from fixing their core problem - that they don't really understand performance media. They haven't invested in aligning their business with the advertisers, and now they have a whole staff dedicated to supporting the Arbiter program. More inertia, more time wasted, and no closer to the holy grail. And let's be honest - while I'm picking on MSN, we're talking about pocket change for them. What's $100MM when you're running a $30B business?

I think SDC, Arbiter, Poindexter (and to some extent DoubleClick, Zedo, Falk, etc) are part of the problem, not the solution. They're not helping companies reinvent themselves, they're providing stop-gap measures that just postpone the inevitable. Part of me is glad to see companies like Centrport go bankrupt, and if the space wasn't so hot with VC money we would have seen quite a few other follow in their footsteps. I'm keeping my eyes open for a new, more holistic approach - one that acknowledges that managing media the right way requires some sacrifices and reinvention and helps people achieve them.

Today's NFW

I heard two things today that piqued my interest. Three, really, but let's stick to two for now... I'll save the third for a post some other day.

Thing number one: "XXXXX has no unsold display inventory. They give all the remnant to Overture. They're phasing out all of their network relationships because the networks can't pay enough to compete".

Thing number two: "YYYYY had an unsold problem, but they're giving the leftovers (via an optimization partner) to Overture to monetize and expect the solution to product 1/3 of their overall revenue going forward".

Now, a couple things about this should catch your eye.

First, Overture? With all the hype about Google, it's interesting that Overture's getting the traction here. This validates my theory that Google doesn't understand the enterprise media business particularly well.

Second, an optimization partner? Is Overture's built-in optimization on non-search inventory that weak? And if so, how long until they fix it? I'd be shocked to see this kind of arbitrage continue.

Third, textlinks as remnant/RON backfill on major sites? More than anything, this strikes me as a great example of the froth in the textlink marketplace. I can pretty much guarantee that these aren't delivering very good results for the advertisers... and something has to give. I'd love to see the ROI statistics on this... publishers shouldn't rely on this money to last.

On that note... if you think about it, if you're serving text ads into banner slots and with no fancy targeting they're doing better than standard banners, doesn't it mean that the people designing the standard banners are doing a crappy job? Shouldn't happen... text is just one tool in your belt, and you should be able to make it do much more for you.

AdWords content targeting

I read an interesting blog entry today about AdWords that reiterates many of the problems that Google will have in trying to move beyond search.

While it's fairly obvious to those of us in the industry that AdWords is only effective for search, I decided to run a test campaign in order to share the results.

Search:  1045 clicks, $0.53 CPC, $7.53 per conversion
Content: 1129 clicks, $0.52 CPC, $32.17 per conversion

Fascinating, right? With about the same number of clicks at the same CPC, I paid 4x more per conversion on content than on search, even though my average position was twice as high.

So it's broken - but how can it be fixed? Clearly, Google has to offer some kind of CPA optimization. That's going to be very difficult for them since the entire CPC model works because CTRs are usually in the 0.5% range. Conversion rates are 10-1000x lower, meaning that they would have to run at least 1000 clicks per advertiser before starting optimization. It's a hard sell to the small advertiser to have to spend $500 before getting any results.

I think the bottom line is that context isn't sufficient outside of search. You have to tie in other variables to try to account for the fact that the user isn't actively looking for what the advertiser's selling - and that's going to be impossible for Google to do without rearchitecting their system.

Fortunately for Google, the only players that understand how to do this are the ad networks, and they're not getting the hype (FastClick's IPO included) that the context/search players are. I expect that by the end of 2005 we'll see at least one network that figures out how to incorporate context into a multivariable optimization scheme and can consistently beat Google's results outside of search.

However, Google (and other pure-context players) will continue to attract small publishers, as their algorithms don't require any knowledge of how site visitors perform and behave - they just need to slurp the pages. That's a challenge for networks, especially FastClick, which have many small publishers. I prefer the positioning of Advertising.com and Right Media, whose optimization technologies allow them to provide a value proposition regardless of how aggressive Google (or FastClick) gets in paying out to publishers.

While Google doesn't disclose their payout information, I wouldn't be surprised to see FastClick increase their standard payout to 75% in an attempt to gain marketshare. This will be extremely effective for small publishers, but without better technology FastClick is in the same boat as Google - they just can't deliver for quality direct-response advertisers. However, since FastClick doesn't have a search business to fall back upon, they're essentially screwed. More so if the rumors of an attempted acquisition of DoubleClick are true - they don't have the managerial experience to handle DCLK, and any such takeover is likely to be resisted fiercely by management (and investors).

It's going to be a very interesting year. I'm looking forward to watching the fallout as advertisers are priced out of search.

Why Tacoda's Behavioral Network can't succeed

There's been a lot of hype recently about behavioral networks. Tacoda's in the process of launching their network (ClickZ - Tacoda launches behavioral network) and other players in the space are right behind.

Tacoda is a technology company that helps publishers mine data to create behavioral segments. They've spent the last few years convincing publishers that their technology will create additional value from this data, a unique resource that will drive up CPMs.

Yet now they're launching a network, which in essence gives advertisers access to these same segments. This poses a channel conflict with their technology customers. If advertisers buy behaviorally-targeted users through the network, there's no reason to go to the publisher - which means there's no reason for the publisher to buy Tacoda's technology. Moreover, if Tacoda extends the network beyond its clients, there's no incentive to be a technology client at all.

So the technology clients - all major publishers - are doing the logical thing: charging Tacoda money for access to their behaviors. By doing this, they make sure that they get paid for their unique data regardless of who sells it. Makes total sense, but it ruins the cost-efficiency necessary for Tacoda to scale the network. Moreover, they'll sell these behaviors to anyone that wants them, so Tacoda has to compete with networks that have deeper pockets, strong agency relationships, and many non-behavioral deals to show when a particular user isn't in any segment.

My read on the situation is that Tacoda will be gobbled up in the next year. Their brand will be assimilated by a network vying for credibility in the behavioral space, and the concept of a purely behavioral network will disappear with them.

How networks cheat, lie, and steal

Imagine you have these four buddies - Bob, Sam, Dan, and Joe - who want to help you sell your famous ice sculptures. Every week, you send each of them 5 sculptures to sell; at the end of the month you get a check from each of them and a note.

Bob's note says, "Sold all 5 sculptures. I'm not going to tell you how much I sold them for. Attached is a check for $1250."
Sam's note says, "Sold 3 sculptures for $500 each! Here's your 65% cut - $975."
Dan's note says, "Sold all 5 sculptures. I won't tell you how much I sold them for, but here's your flat fee check for $1000 ($200 each)."
Joe's note says, "Sold 3 sculptures for $500 and the other 2 sculptures for $200. Here's your 65% cut - $1235."

So how much inventory should you send each of your buddies next month? It seems reasonable to calculate how much they sold each sculpture for.  Dan's easy, because he'll pay you $200 no matter what. Joe and Bob both sold all of the sculptures, for around $250 each.  Sam sold the sculptures for the most money ($325) but he didn't sell them all.

Your best bet is to give the inventory to either Joe or Bob. If you give the sculptures to Joe, there's no guarantee he'll make top dollar for every sculpture, but at least you can be confident that you'll know what he sold and you'll get 65%. Bob might pay more, but what happens if your sculptures start selling for more money? Will he share the profits? Without full disclosure, you can't be certain.

So how does this relate to online advertising? Each of these scenarios parallels a major ad network.

Bob is like Google. He'll pay you decent money, but you won't know where it's coming from.

Sam is like FastClick. He'll sell each unit for a lot of money, but he won't sell everything - and worse, he won't tell you what he didn't sell. In fact, he'll act like he's the best, since his average price per unit sold is so high- and a lot of people fall for that.

Dan is like Advertising.com. He'll give you a flat fee, so that you don't have to micromanage him, but he'll take a larger cut in return for taking more risk.

Joe is a transparent rev-share network. He'll do his best for you, and he'll tell you what he's selling - and what he's not.

I have real-world examples of each of these from conversations I've had with publishers about how they're divvying up their inventory. When I've explained how FastClick hides its true effective CPM, they've been so mad that they've pulled tags. I think that's a good thing - it's time to punish the companies that won't be honest about how they're selling your product.

(but what about daisychaining? good question - I'll address this in an upcoming post)

About the Ad Inquisitor

The online advertising industry is like the Wild West. We have the first wave of settlers - think DoubleClick and its ilk - that have managed to scrape through a few hard winters. There's a growing stream of new settlers (textlink advertisers and bloggers), many looking for the "gold" in them thar hills. And of course we have outlaws (spammers, spywarers) who are willing to pillage the newcomers, not to mention the shopkeepers (Google) that make the real money from the chaos.

In the harsh reality of this new frontier, I plan to try to expose the snakes and the snake oil. If in the process I enlighten or enrage, I suppose that goes with the territory. And while there's no superheroes in this postmodern tale, I did think it would be fun to choose an exciting nickname. So without further ado, I introduce to you: the Ad Inquisitor.

(and yes, I did recently read Kavalier and Clay. quite a good read)