Intranet search comes with its own set of challenges. This year I’ve seen a number of issues arise with search implementations in intranets. Part of this lies in the fact that, as always with search, it isn’t just a technology issue.
In fact, on the technology side we’ve seen signs of progress. The 2019 edition of the SharePoint Intranet-in-a-Box report from Clearbox Consulting showed intranet vendors paying serious attention to improving the search experience. These vendors face some constraints as to how far they can can enhance the experience as a result of using SharePoint as a platform.
Life is much easier for vendors with their own platforms, for example, Interact, which uses Apache Lucene/Solr; Sorce, which uses dtSearch and Colibo, which uses Lucene/Elastic. But even with these improvements from the vendors, challenges remain.
Designing the Intranet Search Experience
The relationship between browsing and searching was first covered by Marcia Bates in a landmark paper published in 1989. This presents a challenge to intranet architecture design as the traditional card sort/tree test approach cannot accommodate a discovery route which switches between browse and search. All too often I’ve seen architecture specialists look aghast at tests which indicate users want to use search to progress their discovery process. Instead of seeing it as an opportunity to integrate searching and browsing, they read it as a criticism of their architecture.
Usually the search box is small and hidden away in the top right hand side of the screen rather than placed front and center. In addition, search results may be presented without a URL. Including the URL allows users to identify an area in the intranet which is rich in content but which they may be unfamiliar with, to support further browsing.
When initially assessing an intranet, one of the elements I look for is how easy it is for users to provide feedback to the intranet team on search performance. For browsing, click logs can be quite helpful in assessing the success of a user journey — but that isn’t the case with search. At a minimum, some form of icon should be present that people can click to measure their satisfaction. And ideally there should be a feedback form to the search manager. There aremany potential points of failure in intranet search, as is well illustrated by a schematic developed by Sam Marshall.
The challenge here is that for most organizations a full-time intranet manager still seems like a luxury, let alone a search manager. Intranet product vendors are clearly responding to increased expectations of search or they would not be putting in the level of investment they have into the technology. What’s the point of having these capabilities and then not making the investment in the intranet team to optimize the performance?
Things become very complicated when it comes to multiple language search. Recent research from IntraTeam’s intranet benchmark service indicates 64 percent of surveyed organizations only operate in a single language. At a recent IntraTeam enterprise search community meeting (which was carried out in English) I asked how many attendees had employment contracts in English. Not a single hand raised. So while English may be the default corporate language, the intranet is a core repository for content that is local in scope and language.
That has a major impact on search, not only in making sure that all language packs are installed, but also in how to present results in multiple languages (or even in two) to the user. Should the intranet present results in a single ranked list, display them in SERPs that are tabbed by language, present them in blocks of language, or in two or more columns? There’s good research indicating which option users prefer, but intranet managers (and Microsoft) seem to be unaware of it.
Another aspect of multi-lingual search is the extent to which the metadata tags are consistently applied in each language. These tags may well be concepts (‘marketing’) and require a good knowledge of the specific language.
When it comes to metadata, a particular challenge arises with defining expertise in multiple languages so the organization can make good use of its employees’ knowledge. An optimum combination of taxonomy, thesaurus and controlled terms is essential to thoroughly support expertise finding. As an example, I am an information scientist, but neither Google or Bing can come up with a translation of this concept into other languages. The results usually place it as something approaching a computer scientist, which I am certainly not. While this may seem a special case, just try putting your skills and knowledge into either of these translation engines and ask yourself if the words offered really match your own assessment of your skills.
Towards an Intranet Search Strategy
As we head into 2019, it’s as good a time as any to look at your intranet strategy and consider whether it is fit for purpose. In particular, if you operate outside of countries which have English as their national language then you are certainly going to have content in at least two, if not more languages. Does your search functionality provide equivalent levels of performance in each language? If not, why not?
Martin White is Managing Director of Intranet Focus, Ltd. and is based in Horsham, UK. An information scientist by profession, he has been involved in information retrieval and search for nearly four decades as a consultant, author and columnist.