I’m somewhat exasperated by those who see enterprise search user satisfaction as an easily-solved  technology problem. The reality is that enterprise search is unlike any other aspect of search and it is an excellent example of a wicked problem. All the evidence suggests that perhaps 40% of employees are dissatisfied with their internal search application even though the same percentage regard being able to find information being of the highest importance. In this post I want to go back to basics so that we all appreciate the challenges of delivering a superb user experience.

Massive amounts of structured and unstructured content, very little of it curated

In all other search applications, such as web sites, intranets and e-commerce, the content is highly structured by people who have an interest in it being found. In enterprise search employees are not incentivized to make content findable by creating good document structures and adding in metadata. The business objective is productivity and not findability

Often multiple applications in a federated search implementation

Technically it is not difficult to run a query across multiple applications to create an enterprise-wide search. However, the user interfaces are then a nightmare to comprehend. Have you noticed that when vendors promote their federated search implementations they never present the UI options that are available.

Multiple content languages and inconsistent language skills of users

One of my clients had 75% of its enterprise content in English, but only 25% of employees had English as their mother tongue. That makes for some serious difficulties with creating queries, coming up with synonyms for query expansion, reading snippets in fractured English to determine relevance to the query and then reading through the document.

Security trimming plays havoc with relevance

Effective security trimming is sine qua non in enterprise search, but both early and late binding present a host of problems. Equally challenging are the impacts on search satisfaction. Two colleagues may find different documents highlighted as relevant because (not that they know it) they have different security profiles. On seeing what a colleague has found the innocent party then complains that search is broken and of course trust in the search will be broken.

Employees have a substantial about of information pushed to them by other enterprise applications

One of the myths of enterprise information management is that users spend their day looking for information. The reality is that information is being pushed to them by other enterprise applications (ERP, CRM, HR) and of course email and social media.

Search is therefore ‘additive’ and not from a zero knowledge base

Research back in 2012 showed that enterprise users were experts in their domains, and that as a result of information being pushed to them they were likely to have most of the relevant information and documents. What they wanted to undertake were searches to extend what they had accumulated. The assumption of search vendors and CIOs is that all searches are being carried out as though the employees are totally ignorant.

Employees have multiple role and responsibilities, and therefore information seeking requirements

The most difficult aspect of enterprise searching is working out why a search is being conducted, known as the search intent. Creating personalized results on the basis of past searches or documents that have been written fails to appreciate an enterprise search environment where (especially now and for months to come) employees are taking on new responsibilities as colleagues are made redundant and are performing multiple roles. Presenting enterprise search as a technology for mind-reading is not helpful.

Enterprise search is not based around business processes so it is very difficult to assess whether a search failure is a content issue, a query issue or a technology issue

The failure modes in enterprise search are many and varied. When an HR application fails to track down an employee there are only a few possible failure modes. When enterprise search fails to track down an employee and all they have written then tracking down why is a very complex exercise.

Largely managed by IT on ‘technical performance’ (uptime/traffic) metrics with few (if any) search specialists

Because every search user is an individual what they are looking for is individual and usually business-critical. Only with a great deal of attention to search logs and a range of feedback channels can the patterns of search success and failure be defined and addressed, and that takes a blend of a good knowledge of the technology, of the nature of the content being indexed and searched, and the language of the business as far as jargon, synonyms and abbreviations are concerned.

An assumption is that search is intuitive, so no training and no mentoring available

In the case of most enterprise applications employees are trained in how to use them and often their ability is subject to certification. There is no evidence that search is de facto intuitive, but a substantial amount of how important training is in ensuring that search users are satisfied with the performance of the application.

I should add in closing that there are in fact more than 10, but it seemed a good place to stop….for now!

Martin White