The Search Learning Curve

I spoke on the ‘The Alternative Search Inventory’ panel at the Piper Jaffray Online Advertising & Search Symposium together with Tim Tuttle of Truveo and Erik Matlick of Industry Brains.

Another good panel at the same event was ‘Search Destinations and Technologies’, with Tim Cadogan from Yahoo, Tim Armstrong from Google and Arkady Volozh from Yandex. One of the more interesting observations of the panelists was that the average query length in web search continues to increase. In other words, web searchers are getting better at searching, using more terms to drill down more deeply.

There’s a similar trend in vertical search. The average job search query length on Indeed, for example, has increased by 14% in just 10 months; from 2.2 words in January this year to 2.5 words in November (excluding location terms). As job seekers become familiar with using Indeed, they are quick to learn how to drill down through millions of jobs, to find precisely the right ones in the right place.

A live demo I did at the symposium is a good illustration of the effect of small increases in query length. I asked the audience to give me a sample search query. The first one volunteered was Research Analyst (people always pick their own jobs!) – over 30,000 results. We added a location – New York City – which took it down to about 3,000 results. I asked for more specificity and the audience piped up with internet, which reduced it to about 200 results. So, in this example, going just from two keywords (Research Analyst) to three (Research Analyst Internet), narrowed it by a factor of 13 to yield a very manageable set of results.