Searching Smarter

There's a lot of chatter these days about making online search easier, friendlier, serving more of the public in more ways.

For example:

  • natural language search, where we'll just type (or speak) in plain language
  • using prior search, adjusting future results to more closely reflect our past selections
  • social search, where the preferences of our network of friends will influence our results
  • emphasis on optimizing the very first search result since that's the one we'll want to click
  • and so on...

That's all fine, but do we see an underlying assumption emerging? 

Perhaps a problem (and an opportunity) in disguise?

Do generality and ease of use carry a cost in terms of specificity and performance?

In the opening paragraph of his book Good to Great, Jim Collins wrote

"Good is the enemy of great.  And that is one of the key reasons why we have so little that becomes great.  We don’t have great schools, principally because we have good schools.  We don’t have great government, principally because we have good government.  Few people attain great lives, in large part because it is just so easy to settle for a good life."

The rising tide of information technology raises all boats, but those with better navigation, speed and agility still tend to win the race.

Probably the biggest immediate potential improvement on today's search results would be to run a secondary query over the initial results set. Download those 50 or 100 pages into memory (takes less than a second) and perform a more advanced query on the contents; this time you're able to specify such things as capitalization of product names, specify words near each other in the same sentence, and let's say you want to get back anything and everything that looks like a postal address, but only for your city. 

The scenario above is readily doable, given the right tools, because it's focused and targeted.  But it's not something any of the general search engines will do, or could do, without sacrificing the generality that gives them their strength.

Now suppose you'd like to take each of those secondary results (containing an address in your area) and check whether they belong to a web site which contains your key words of interest on any of it's pages, or maybe get names and addresses from any "about us" or "contact us" pages.  No problem, but that's a topic for another post.

Or email us with an idea of what you'd really like to do...