Thursday, April 07, 2005
Received a couple of mails essentially saying 'you're right, contextual targeting is so over, behavioral targeting is next'.
In response to my post arguing that contextual advertising is likely to be disrupted this year, because it's value prop to both advertisers and audiences is decaying (growth of click fraud, web spam, etc).
This is intimately related to the emergence of vertical search/ecommerce, which partly is due to innovators seeking incremental efficiency gains in search costs, but also partly due to general search value prop decay. It's becoming less economical to use general engines as the web grows, because the growth of the amount of crap (spam, dead homepages, blogs no one reads, etc) is still accelerating. This gap in market space makes verticalization a dominant strategy.
So what's next? I don't think it's gonna be behavioral targeting.
There are many reasons for this. Remember, we saw early (relatively inefficient) versions of behavioral targeting during the bubble. But ad networks on the Net were clearly a winner-take-all market if ever there was one - and Google won.
Why was contextual advertising superior? The economics of behavioral targeting don't make any sense. The benefits are clear for advertisers, but not for audiences. Since behavioral targeting imposes greater costs on users (cost of privacy loss, cost of lost attention from more invasive ads) than contextual targeting does, it must offer audiences greater gains as well.
How could it do this? Well, the most straightforward way is to push ads that consumers want to see. Unfortunately, behavioral targeting can only be marginally better (if that) at this than contextual targeting can. Assuming behavioral targeting can't offer consumers gains, consumers have great incentives to disable behavioral tracking systems - and they do. What this, in turn, means, is that behavioral targeters have the incentive to hide their actions from consumers - it's a classic case of moral hazard. This is how adware/spyware emerged.
So contextual targeting has, and I think will continue to, be more efficient at advertising to consumers the stuff they're searching for.
What this means is that behavioral targeting will continue to stay locked into the (profitable) niches that it's been successful in: serving ads in places where consumers spend a lot of time (and so can be costlessly tracked). Think NYT. In other words, behavioral targeting will pick up where contextual targeting stops. You search for 'Ferrari Dino' (and see contextual ads), and then visit a car review site (and see behavioral ads). This is the thinking behind the funding of plays like Revenue Science from fairly heavy hitters.
So what will disrupt contextual ads? Well, the important bit to note is that the bits of the value prop which are decayed are not contextual targeting: they're the open access mechanism, which enables click fraud, and the general search mechanism, whose efficiency is growing less slowly than web spam/noise growth.
I've talked about insurance and/or risk-sharing mechanisms before - my money's on this reducing the costs of click fraud: solutions that allow the risk of open-access contextual ads to be massively shared or insured. I expect this solution to be implicit, not explicit.
As for search efficiency, there are two solutions. The first is vertical search. The second, I've argued, is going to be horizontal. The first glimmerings of this are beginning to emerge in RSS aggregation.
The natural trajectory is to extend standards like RSS across media, so you can have feeds of all the stuff you like aggregated horizontally. This allows profile-based targeting, which I think has the potential to be the aforementioned disruptive tech - it can be superior to both behavioral and contextual targeting on value-driving attributes, but inferior in others.
This is why FeedBurner's recent B round is a good bet - though I'm not sure whether this line of analysis fed Feld's investment rationale, which equates RSS to SMTP, or Dick's strategy, which centers around being a simplicity magnifier.
Awesome post. Your blog is awesome for the type of econ it brings to the net's attention, I think. Fascinating to a non-economist like me.
BTW, with your attention scarcity model, you might be interested to know that your posts on Apple and iTunes where what brought my attention to your blog via Macsurfer.com. And I don't have time to invest in the other econ blogs to which you link.
there are two reasons that you haven't yet seen the impact of behavioral targeting: 1/ it's hard to do as a technical matter, much harder than contextual targeting, 2/ only a few players on the web have the kind of behavioral profile on users that you need to do it well (here i mean broad sites like yahoo and msn, but not google, which is only about search), and 3/ yahoo/overture was clinging for too long to the notion of fixed placement paid search listings, which is 100% keyword (context) driven. in fact, when yahoo and google really get going on a combination of contextual and behavioral targeting (applied to the same presentation of results) the power of this will be clearer ...