Ad Inquisitor

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.