Many start-ups are trying to usurp Google, over 1,000 in fact. In my view, most of them are likely to fail, because they don’t really add that much value. Here is why I think that.
Google (and live.com and Yahoo! Search, and ask.com) all fundamentally answer the same question a user has: “Where can I find…” This is very important. With Google, you fundamentally ask the question of “show me the most relevant websites where I can find the following keywords”. The vast majority of search start-ups out there do the same thing, to answer where you can find information. They just try to do it better. Frankly speaking, this is a lost battle. In order to take any significant market share, you would have to do significantly better than Google. That is a very hard thing to do. So, why bother?
I think a much better way of looking at this space is not to address the where question, but a different one. For example: “How do I…” Here, the user doesn’t want to know where to find information on a keyword, but wants an answer to a specific question. For example how to file a tax return. There is one company in this space that is doing well, it is called Mahalo. Jason Calacanis at Mahalo seems sometimes amazed about how much traffic the ‘how to’ pages at Mahalo get. To me, this makes a lot of sense. Human editors are very good at processing slow-moving content where insight is required. ‘How to’ matches that very nicely. Even I started using Mahalo for ‘how to’ questions, it actually works quite well. Have a look at the compete stats to see how much traffic they get overall.
I think there are many other questions search engines can answer. For example, you could ask “How good is X”. For example, I am convinced a review aggregation search engine would do extremely well. I cannot remember how many times I have typed in the name of a product and ‘review’ just to get to all the crap sites out there that list where I can buy stuff, when all I want is to see all the reviews on a product, regardless of where they originate from. How hard can it be to aggregate reviews from some 1,000 leading sites and to display that on a central site with back-links to the original sites? Some start-ups like testseek reevoo or buzzillions are going in this direction, but I think they don’t go far enough.
UPDATE: I forgot to mention the “What is…?” question. That has been covered by Wikipedia. Very difficult to compete with that.
I am sure there are many other examples out there that can work. In my mind, when you try to build a new search engine, don’t ask yourself how you can make the where answer better, rather ask what other questions would be really valuable that Google cannot answer.
The best way to compete is to not compete. Particularly not with companies like Microsoft or Google.