I continue to be disappointed by the continuing use of a figure of 2.5 hours a day for the time that employees spend searching for information. This figure appears quite frequently in business cases for investment in search applications on the basis of a ‘time saved’ calculation worked up around average salary costs. This data item first appeared in a research paper from IDC in 2001. One of the authors of the paper was Sue Feldman, who without doubt has had a distinguished career as a search industry analyst. Sue went on to establish Synthexis and write a book with the intriguing title of The Answer Machine.

It is worth reading the 2001 paper on The High Cost of Finding Information in detail. The enterprise world was different in 2001. “Preliminary research suggests that approximately 35–50% of the information available within an enterprise is not centrally indexed. This information resides in databases and desktop or notebook computers”. The focus of the report was very much on intranet search at a time when SharePoint had only been on the market for a couple of months. The authors go on to state that “We use a general estimate that the typical knowledge worker spends about 2.5 hours per day, or roughly 30% of the workday, searching for information. This number also needs to be adjusted to reflect the circumstances of each specific enterprise. IDC believes the number represents a general average of time spent searching based on the ubiquity of intranets within organizations.” No statistical basis for this estimate is given in the paper. All that Sue Feldman and Chris Sherman were doing was developing a scenario highlighting the potential cost of finding information, the title of their analysis. They were not making a definitive statement that would stand the test of time over the next seventeen years.

Professor Pia Borlund (Oslo Metropolitan University) has been researching interactive information retrieval applications for over a decade. In 2012 Professor Borlund together with Sabine Drier and Katriina Bystrom, published a paper entitled “What does Time Spent on Searching Indicate?”. A very good question. The authors reviewed all the published research on the topic and undertook some experiments in both a laboratory setting and inside an organisation. In the research review there is no reference to the IDC paper. Interesting!

More recently Professor Karvo Jarvelin (University of Tampere, Finland) and Miamaria Saastamoinen have used computational ethnography to gain direct experimental data on the time that people spend searching, and this was also an element considered in a very recent study on professional search by Tony Russell-Rose, Jon Chamberlain and Leif Azzopardi.

When talking about the time spent searching the initial challenge is in deciding when a search has been completed and the clock can be stopped. In enterprise search there are often multiple ‘sub sessions’ as users look at documents and then revise their queries. The evidence from the research noted above is that search sessions might take anything from 20 seconds to many hours, with so many variables involved that an ‘average’ time is a meaningless statistic. It might take just a minute or so to identify an important market research document and then several hours reading through it and analysing the information. Is the ‘time spent searching’ a minute or several hours?

Another issue to think about is how realistic is it that nowadays any knowledge worker could find the time to spend 2.5 hours on searching. Since 2001 the typical number of workplace emails has increased by a factor of ten, and probably the largest single block of time in a working day is allocated to email management. If knowledge workers really are spending 2.5 hours a day on searching for information then in effect they are working a four-day week, and no manager is going to accept that state of affairs. Another feature of work in 2018 is the extensive use of team working. Teams are a very effective way of sharing information. One person with a key piece of information will save everyone else in the team from having to undertake individual searches.

The 2.5hr discovery time was probably a decent guess in 2001 and was used at the time to make a very plausible business case for search. All the recent thorough experimental evidence confirms that the “time to complete a search” is a very complex metric and basing a business plan on a search time/salary calculation alone is not the slightest bit credible. If you want to make a better business case then come and talk to me or any other member of The Search Network.

Martin White