chameleon hiding in the leaves
PHOTO: Juan Carlos Fernández Rodríguez

For years now I’ve sat through conference presentation after conference presentation given by search managers. And no matter how many I sit through, I never hear what I'm most interested in: the usage data. Instead these managers share limited details, some heavily edited results pages and something about the technical architecture. Unique users and queries per day are hidden behind the corporate security blanket. 

You would think a new search application would be a reason to talk about the significant increase in adoption and use. And surely discussing core search numbers isn't giving commercial intelligence to competitors, is it?

The cause for the secrecy is quite different. Too often, usage data is not disclosed because not only are levels historically low, but they also have not substantially increased since the new technology arrived. How embarrassing for those who tout it as a solution employees will use every day of their working lives!

Over the last 10 years I’ve been involved in a dozen enterprise search projects. It doesn't take long after starting on the user requirements to find the problem is not information scarcity but information overload. The picture of an individual employee sitting at their desk desperately using a search application to find information is false. A more accurate picture is of the employee sitting in a stream of information being pushed at them by perhaps a dozen enterprise applications, as well as email and social networks. 

Given how organizations continue to profit and succeed in spite of the scarce attention paid to search suggests to me that a majority of employees do not see lack of information as a problem.

Who Needs Search?

One reason why employees are not sitting at their desks bemoaning their information-deficient state is they are members of multiple teams. An important function of these teams is to share information. It only takes a meeting or two to assemble enough information assets to sustain the team for many months (and usually circumvent security restrictions in the process). 

The result? Employees receive regular injections of information from enterprise applications which is then topped off by information exchange with peers and colleagues. Who needs search?

Related Article: Digital Workplace Success Relies on Strong Search

Zipf's Law in Action 

The diagram below plots the number of times a word was queried in the course of a month in a global high-tech manufacturing company with around 75,000 employees. This is real data from the corporate search logs. The x-axis indicates the rank order of every query. The inflection point is around the 10th most highly ranked query. After that you'll notice the very long tail. Every one of the top 20 (in fact, most of the top 100) queries are people hunting for applications, because they need to access 15 to 20 different applications on average every month. 

Case in point: if you are working in the enterprise resource planning application and want to book a room, the fastest way to do so is to search for the room booking application because it will recognize you when you find it. 

query count vs query rank


A very arbitrary rule of thumb is the top 20 queries will account for 20 percent of all queries. This is an example of Zipf’s Law, which arises in any large collection of text. Your log analysis may show an inflection at a different point but the basic shape will be the same.

Related Article: Enterprise Search in 2018: What a Long Strange Trip It's Been

So Why Do We Need Search?

Our high-tech manufacturing company above saw a total number of around 200,000 searches per month come in from 50,000 unique users. This means that on average employees were undertaking a total of four searches a month, and that includes the application-seeking queries. 

Employees turn to search applications when all other channels let them down. So if search then fails, they really do have a problem. And the low level of use means most employees are not using the search application often enough to be familiar with it. Yet training in search is a rarity. 

Miles Kehoe has suggested that metadata on location and role could help optimize a search experience, but in the case of this company, searches for "core" technical terms were coming from over 20 different countries and (as far as we could work out) many different job roles and departments. Finding cognitive patterns with so little contextual information is going to be a significant challenge.

Related Article: Relevance Engineer: A New Profession in Search of Candidates

Search Isn't 'One-Size-Fits-All'

Search remains an invaluable application — but it is not a universal application used by most employees every day. Search satisfaction is low, but that is a combination of technology issues, content quality issues and a lack of search skills. Employees use the search application as a last resort. Where can they turn when the last resort doesn't work?

Once again, we return to the need for a strong search support team. Only with this team, and a search manager with a very deep understanding of the information-seeking and information-searching needs of employees, will search succeed where all other channels failed.