March 27, 2007 Leave a comment
Skrenta posted the first part on how to take Google on. It it a very interesting read and I would urge anyone who is interested in getting into search to read this article. There is a lot there even if you are not interested in taking on Google.
Some excepts with comments:
A conventional attack against Google’s search product will fail. They are unassailable in their core domain. If you merely duplicate Google’s search engine, you will have nothing. A copy of their product with your brand has no pull against the original product with their brand.
It would be really hard to disagree with this, taking Google on directly would be a little like the Charge of the Light Brigade, perhaps poetic and/or glorious, but certainly futile and definitely murderous. The Google brand is very well established and very strong.
You need to position your product to sub-segment the market and carve out a new niche. Or better, define an entirely new category. See Ries on how to launch a new brand into a market owned by a competitor. If it can be done in Ketchup or Shampoo, it can be done in search.
Agreed here too, you need to focus on a very specific area. Once you have established your brand in that area, then you can move into other areas. Remember that Google was once just a search engine, now it is an ad network, an email services, etc…
Forget interface innovation. The editorial value of search is in the index, not the interface. That’s why google’s minimalist interface is so appealing. Interface features only get in the way.
Forget about asking users to do anything besides typing two words into a box.
Users do not click on clusters, or tags, or categories, or directory tabs, or pulldowns. Ever. Extra work from users is going the wrong way. You want to figure out how the user can do even less work.
Your results need to be in a single column. UI successes like Google and blogging have shown that we don’t want multiple columns. Distractions from the middle with junk on the sides corrupt your thinking and drive users away.
The average consumer just wants results fast, admittedly there are user who will want to have lots of tools available to them to do a deep a search as possible, but the mass market wants to enter two terms and get results, that’s it. Is more is to be done, then it needs to be done out of sight of the user and independent of the user, so somewhere in the 2 seconds it takes to run and return the search results. By all means gather information on which results the user views so you can personalize future searches.
Additional user interface widgets just don’t get used, period.
Your core team will be 2-3 people, not 20. You cannot build something new and different with a big team. Big teams are only capable of duplicating existing technology. The sum of 20 sets of vision is mud.
Search is more about systems software than algorithms or relevance tricks. That’s why Google has all those OS programmers. You need a strong platform to win, you can’t just cobble it together as you go like other big web apps.
Based on past figures, I would estimate that Google has between 450,000 and 500,000 machine now, of which 20%-25% are dead. That still leaves a lot of machines. With those kinds of numbers, it makes perfect sense to have a specific Linux distribution for internal use (Goonux anyone?) I would imagine that this is a bare bones distribution containing only what is needed from Linux, overlaid with whatever tools you need to manage the machine.
Do not fear Google’s vast CapEx. You should wish maintenance of that monster on your worst enemies. Resource constraints are healthy for innovation. You’re building something new and different anyway.
Having that many machines is both a blessing and a curse. On one hand you can bring a LOT of hardware to bear on any problem you wish to solve, on the other hand that is a LOT of hardware to pay for, manage, maintain, dispose of, etc…