Search has to be intuitive. Really?
I’ve lost track of the number of times I’ve seem the words “Needs intuitive search” in either a set of personas or in the RFP document for a new search application. All too often the reason for asking for intuitive search is because all the budget has been spent on the technology and there is nothing left for a skilled search team. There is a common extension of the requirement along the lines of “Needs intuitive search like Google”.
Let me start with Google. It comes as some surprise to many to learn that Google is only indexing perhaps 5% at most of the internet, and probably less. Searching the deep web is very challenging, and consultants such as Karen Blakeman are running highly successful courses on the topic. At least inside an organisation you will know exactly what has been indexed! Even 5% is an immense amount of information and getting the best out of Google requires an excellent knowledge of some of the ‘tricks of the trade’. Again, there are many courses on searching as people begin to appreciate the breath of knowledge needed to achieve even a close to optimal search.
A factor that is relevant to Google search but even more so to enterprise search is the extent to which domain knowledge has an impact on search strategy development, relevance assessment and search satisfaction. People in the same role, the same project and with the same experience will still approach a search in different ways, especially with the way they construct their query, and their use of filters and facets. Recent research indicates that Boolean query management is not dead and indeed the 2d Search application may be taking it to a new level. Many of the services designed for professional search offer a range of user interfaces, recognizing (for example) that users will need different levels of information about a potential result in order to evaluate its relevance. Thomson Reuters CheckPoint is just one example. If these services designed for professional searchers need a range of interfaces then it is likely that users with even wider range of domain knowledge in an organisation will benefit significantly from carefully designed and tested user interfaces.
The initial recognition of both the power and the complexity of the human-computer interface arguably dates back to Douglas Engelbart and his 1962 paper on Augmenting Human Intellect: A Conceptual Framework. Since that time probably thousands of research projects and papers have been published on how to optimize search user interfaces. Marti Hearst, Tony Russell-Rose and Ryen White are just three authors who have written excellent books about the process of searching and what makes for an effective search user interface.
Of course one of the challenges of search management is that the moment you have ‘optimised’ a UI and released it, users will immediately push it to the limits and quickly build up a list of benefits and weaknesses. Changes of direction of the organisation will also have an effect on the types of search that are being carried out. Constant testing and development is essential.
In short trying to achieve an intuitive search interface is a never-ending process. If Google has not solved the problem after 20 years of research then it is highly unlikely that there will be an enterprise-level solution any time soon. Will AI and machine learning provide solutions? Possibly, but how long are you willing to wait?