“Vertical search engine” is an oxymoron. Nevertheless, a new report from Jupiter Research uses a bad analogy from the development of TV to predict that the search engine industry will go vertical. The report is summarized on the SearchViews blog and there is a press release on BusinessWire. It won’t happen, says I, because broad search engines which span the whole web are what users want and need – even for extremely esoteric searches.
[Disclosure: I didn’t buy the full report from Jupiter. I am relying on their summary and the summary on SearchViews in arguing with it.]
Jupiter points out, correctly, that both viewers and advertisers benefit from the proliferation of vertical TV channels. Users get the exact content they want; advertisers get the exact audience they want. Google, Yahoo, and MSN are like the old ABC, CBS, and NBC, they say. Soon there will be a proliferation of targeted search engines like the 500 channels of satellite TV, they predict, and advertisers will be able to focus their increasing buys of search engine exposures on exactly the engine whose users have exactly the demographic they want.
Wrong! The analogy is broken:
Each TV channel can only offer 24 hours of content per day. Obviously, more content requires more channels. The power of today’s search engine is that they offer the whole web and their reach grows as the web grows. We don’t need more search engines to cover more content the way we needed more channels.
Because they can only broadcast one show at a time, “mainstream” channels avoid exotica. The web doesn’t suffer from bandwidth limitation. Exotica, minutia, trivia, and fascinating detail unheard of on any TV channel are as available as “mainstream” information through any search engine searching the whole web.
Except for regionalization, a TV channel offers only one ad at a time to its whole audience so you need a different channel to reach a different audience. In theory, the combination of ads that pop on each of the billions of search pages returned each day could be unique. Advertisers already get a degree of targeting impossible on television by the use of general purpose search engines. They have a good chance of reaching the demographic they want by targeting the search terms they want to associate with. Can this be better? Of course, and it is improving. But the answer isn’t a proliferation of narrowly focused search engines.
Even if vertical search engines were better for advertisers (which I don’t believe they are), they wouldn’t catch on because they’re not good for users. Do you really want to have to guess which search engine crawls the part of the web which has the information you want today? Do you really want to learn a different search engine UI for each category of information you want?
Looked at another way, what do you gain by having your search engine NOT index part of the web? Nothing, is the answer. What do you lose if the search engine only indexes part of the web? A lot. Suddenly you are dependent on whoever makes the range decision for the search engine seeing the world the way you see it. You would have to understand what each search engine meant by categories in order to decide which search engine to use for a particular search. And, it might well turn out that the category you are interested in isn’t covered by any vertical search engine.
If the users won’t leave the great horizontal search engines for vertical ones, it doesn’t really matter if advertisers wish they would (Jupiter doesn’t seem to understand this basic marketing fact – at least in the summary of the report they published). However, advertisers wouldn’t benefit either. Targeting by key word already gives more selectivity than being limited to “only” 500 channels. In various ways - some of them very legitimate, some of them not – actual information about the actual consumer can be considered along with the search words that indicate the consumer’s current interest and targeting can be incredibly precise.
On TV, an advertiser who chooses the Golf Channel to sell golf clubs is missing all the golfers who happen to be watching other channels at the moment. This is often the right tradeoff to make given the limitations of TV. But these limitations don’t exist in a horizontal search engine. I can reach everybody who uses a golf word by buying those words. Just as on the Golf Channel, I don’t pay for those who aren’t interested in golf. But, better than the Golf Channel, I get every customer of the horizontal search engines who is currently interested in my product.
I am not arguing that users won’t congregate around special interest content on the web just as they do around TV channels. Nor am I arguing that an advertiser should ignore direct placement (or indirect placement through a search engine) on sites with specialized content which draws distinct demographics. Specialized sites will continue to have specialized search interfaces for their own content – particularly if the content itself is inherently structured. However, it is interesting to note that many sites have abandoned their internal search capability and, instead, just enable Google or other horizontal search engines on their site. My blog and most others do this. Much more importantly, so does Wikipedia which offers a choice of Google or Yahoo.
It is also true that communities form around common interests on the web. These communities usually have some mechanism for referring items of presumed interest to each other. The web sites of these communities are another place for targeted advertising. However, even here, I suspect that the capabilities of the horizontal search engine will be used in most case to serve up the targeted ads.
When the Internet was first coming into popular use, many people argued that it was too big and too unfocused. They argued that there would continue to be specialized nonWeb networks like AOL and MSN were at the time. That was wrong. Those specialized networks have had to embrace the breadth of the web in order to survive. Engineers argued that there would continue to be a proliferation of non-TCP/IP networks to meet special needs. They were wrong.
The fallacy of these arguments and of Jupiter’s argument for vertical search engines is that they ignore Metcalfe’s Law. Even though a specialized network can be tailored to special needs, it is extremely rare for the value delivered through that specialization to exceed the loss that comes from having a small network rather than a large one. The value of a network is directly proportional to the square of the number of endpoints. The total added-together value of ten networks of one hundred endpoints each is only one-tenth the value of a single network combining all one thousand endpoints!
My prediction is the opposite of Jupiter’s. The web isn’t going to become more like TV. TV is going to become more like the web. The next generation of DVRs with an ability to pull content from anywhere on the web and to serve specialized ads will make TV channels increasingly obsolete. Programs will matter but channels will disappear as an unnecessary vertical construct in a horizontal world.
There have been lots of comments and other blogs on horizontal vs. vertical search engines since I wrote this. I learn from some and argue with others in The Vertical Search Engine Debate.