I was fortunate to be invited to talk at Enterprise Search Europe in London this week, which gave me an excellent two-day overview of the state of search. It was an impressive event, combining practitioners, academics, and visionaries, with detailed, blow-by-blow presentations from large corporations on how they switched the enterprise search system to a new platform (Reed Business Information), or how they implemented search (Airbus Industries), alongside descriptions of small-scale investigations of how search works or doesn’t work in SMEs or institutions. For me, the most valuable lessons were often points that were incidental to the main presentations, but which seemed to emerge as common themes. Here are nine of them:
Search is not seen as a core case for investment. Organisations tend to see search as something to be implemented, then forgotten about. To some extent, this is true also of websites: the website owner worries about getting the search right up to the launch, at which point making further changes to search often becomes strangely less important.
Search is not a product, but an ongoing service. It is clear that developing a good search service is an iterative process.
How do you measure improvements in search? It’s trite, but less time spent on search may mean the navigation is more successful. Martin White came up with the wonderful anecdote about the organisation that prided itself on having 9,000 searches on the corporate intranet for the company teleconferencing phone number. What they didn’t realise is that the huge number of searches was because nobody could find the teleconferencing number on the pages about teleconferencing, so they had to keep searching for it.
There is gold to be mined in analysing the searches carried out by users. This conclusion was elegantly demonstrated by Jennifer English from her work with a healthcare organisation. Even though her methodology raised a few eyebrows - she argued for investigating the “long tail” of searches and grouping them by clusters, but seemed en route to forget about the need to disentangle homonyms. Nonetheless, she made a good case for revising the organisation taxonomy based on what users are searching for.
Exploratory search represents a surprisingly important aspect of enterprise search. This came from a survey of the oil and gas industry, but seemed to be echoed by several of the presentations.
Nobody is praised for creating information within the enterprise. Uploading content to the corporate intranet does not get you promoted, even though it may provide real benefits for the company.
Google provides results based on authority and popularity, neither of which are related to relevance. This suggests the kind of search solution provided both internally and externally by organisations should not be slavishly modelled on Google. Hence the next maxim:
Search is not a “one size fits all”, but needs to be tailored.
The most popular search on the Airbus intranet is for the canteen menu. 80,000 users, 200 separate intranets, 4,000 business applications, and what do the users do? They search for what’s on the menu.