“Why can’t our enterprise search be like Google?”

I’ve lost count of the number of occasions I have had this question from senior managers. Listening to a virtuoso presentation from Karen Blakeman as one of the ISKO UK/BCS IRSG Exploring Information Retrieval lectures yesterday caused me to realise that I had not blogged specifically about this question. In my experience talking about the background technology of Google and how it fails to cope with enterprise content is not a good place to start. Try explaining the implications of Google not supporting Boolean search!

Instead, I focus on ‘trust’ and ‘training’.

Let me start with ‘trust’. My initial response is to ask the manager if they have ever made a decision based on a search through Google (and only Google) that might put either their careers at risk and/or the business at risk. With enterprise search this is quite a common occurrence because employees have no other option than to rely on the performance of their search application. Invariably managers cannot remember such a situation and find it strange that I would even ask that question, as of course they would not rely on Google.

The second issue that comes up is the expectation that search should be intuitive, and they find it difficult to use the current application. Listening to Karen it became clear almost from the very first slide that getting the best out of Google requires an investment in training. Karen has prepared a guide to the main commands in Google but this is only the tip of a considerable iceberg. Karen’s seminars on Google are usually half-day workshops. Having attended one of these workshops I can attest to the fact that even with this amount of time you feel that you have only gained an initial overview of Google’s functionality. Moreover, to get the best from this training it is important to use these commands on a regular basis to check they have not changed (Google makes changes without notice) and it is easy to forget some of the nuances of Google search without regular practice.

There is a landmark piece of research from Paul Cleverley and Simon Burnett that indicates the three main factors that give rise to search dissatisfaction are inadequate technology functionality, content quality and a lack of training. The easiest of these to fix is training. There is a classic case study from the Danish Police where putting a simple A4 poster guide to search alongside the coffee machines resulted in a very significant increase in use and search satisfaction.

If I have time I also highlight the differences between the interfaces of the ‘public’ version of Google and Google Scholar, NGram Viewer and DataSet Search, to give just three examples. One UI does not fit all.

I do want to leave on a positive Google note – take a look from time to time at the Google AI blog to appreciate the significant advances in search that Google is pioneering.

Martin White