With internet content growing at a staggering rate,
search engines need to hone their approach to get the right results
and unearth what the user wants.
Typing “woppit” into Google provides 507 results. In Yahoo, 350.
In MSN search, 528. I have no idea what a woppit could possibly be,
or what I would do with one – I just chose a made-up word to see
what came of it. That such a word produces many results is strange
enough – that we get hundreds does not bode well for when we are
searching for something that matters.
Now, if I type something that I am more likely to be interested
in, such as “saving for retirement”, I get 24,400,000 results from
Google, 6,160,000 from Yahoo and 2,551, 184 from MSN. Great –
somewhere in there is possibly something I would find useful. But
where?
Search engine suppliers use algorithms to ensure that the most
apposite results appear closer to the top of the list, but if the
result I want is even in the top 1% of Google’s solution, I still
need to go through more than 250,000 results.
To give me just 100 to go through would require Google to be
250,000 more selective – and would such selectivity guarantee that
I got the matches that I was searching for?
With the continuing growth of web content (75,000 new blogs a
day, never mind anything else) this can only get worse.
The problem is compounded by the use of paid-for placements. The
top selections in many search engines are there because the company
involved has paid money to try and ensure it is on the first page
of searches that might be right for them.
However, most searches are too woolly to create a narrow set of
results, which means that a lot of paid-for placements are
presented wrongly and may push the right solution to page five or
six of the search, generally beyond the attention span of most
searchers.
There are two ways to address the problem. One is to retrain
users so their search terms are more accurate and they understand
Boolean searches, thesauri and taxonomical systems. This is the
most elegant solution but it is unlikely to happen.
The other is to look at how current search engines enhance the
information they supply back to us. Many now allow a “find similar
pages” function, and they all have advanced search capabilities
which are made as easy as possible, but are used by very few
users.
The one thing that seems to be startlingly absent from most is
an iterative search capability, or being able to play with the
existing search results – having got 24 million results for “saving
for retirement”, I may want to narrow them down by refining the
search, or I may want to rank by date, by geographical proximity to
me, or whatever.
The major search engine suppliers have a lot of new
functionality coming down the line – a lot of it based around
multimedia and “revenuisation” of their capabilities (ie, making
money out of a free service). I still use search engines for a lot
of my daily work, but I have less faith in the results day by
day.
There are many things that can be done to help in this way with
little extra technical exertion by the search engine provider.
Being able to enhance results by sorting and refining existing
searches is pretty easy, for one. Other things may need more work,
and no doubt we will see refinements continue to appear.
We no longer want a search engine, we need a find engine. I want
the capability to drill down to apposite results rapidly. I have no
interest in seeing that a search engine can find an order of
magnitude more results than the next one – I want the 100 results
that are right for me. But, like most people, I do not want to
change the way I search. I want a solution that works with me to
get to the desired point – rapidly.
Clive Longbottom is service director at analyst firm
Quocirca