How Sticky Is Your Enterprise Search?
If you manage an ecommerce site, you can probably tell me exactly how many visitors you have, which products or services they look at and which products or services they buy. You likely also know the abandon rate, how long each user stayed on your site, and possibly what part of the world has the highest purchase rate. Chances are good you can predict seasonal trends pretty accurately. And you may be using machine learning, or at least past site history, to direct future users to the most likely pages — based solely on previous visitors’ actions.
Even if your organization doesn’t sell products on your public site, I would guess you have other staff who know a lot about your visitors: the search terms that brought people to your site, the most common path visitors take through your site, and most likely the click-through rate. All of this is simply standard practice for organizations on the web these days. Much of this falls under the umbrella of search engine optimization (SEO), and hundreds of companies and tools are devoted to helping businesses get great coverage on the web.
It's Time to Discover Internal SEO
Sadly, most internal enterprise search teams don’t spend much time looking at user behavior, much less which site pages are popular. Even with authenticated users and known IP addresses, internal teams rarely know which departments, let alone individuals, are using internal site search, or what people are looking for. Some don’t even know what queries are common, which return no hits, and how many searches are abandoned.
I’m sure you’ve heard of SEO. These methodologies help businesses manage how their public-facing web properties perform. Virtually every organization with a public-facing website has some staff responsible for insuring their site content ranks well on Google and other public-facing search platforms. Some companies are also savvy enough to staff a team to optimize their own internal site search, which is especially true when it comes to ecommerce sites.
More recently, people are starting to consider ESEO, enterprise search engine optimization. ESEO means devoting the same time and attention as given to SEO to reviewing activity on your internal enterprise search. The difference between activity on your public-facing sites and your internal sites, though subtle, is simply the scale of the operation.
Related Article: Why Is Enterprise Search So Hard?
Reading the Query Log Tea Leaves
Most of the queries you’ll find are mundane and even predictable. Topics like holidays, vacation schedules and maternity leave policy may generate lots of activity, but you can extract a great deal of learning from your internal site search.
Consider the more serious topics your internal query logs may reveal. If employees in one department are searching for "harassment policy," could that be valuable information? I’m certainly not a lawyer, but my understanding is if a company could — or should — have known about harassment, they could be held liable. Could your search logs be used against your company as proof?
Even without legal risk, ignoring your search activity logs could be costing you money. We recently worked on a project to upgrade site search for a successful ecommerce company. As we reviewed the existing query activity, we noticed a large number of queries for product names and part numbers were producing no results. When we brought this up with the product team, we learned that pretty much all of the "bad queries" were for old products that had been replaced by new product lines. After a simple tweak to the search platform, the site displayed the new product lines for these queries, virtually eliminating the revenue sucking "no hit" response.
Related Article: Buying an Enterprise Search Solution? Don't Forget an Integration Partner
What You Can Do With Context
Using context helps. Ecommerce sites understand this and use machine learning and other technologies — often in conjunction with search — to build internal profiles of site visitors. The good ones recognize a return client, and can make recommendation based on previous site visits — even if that person never purchased any products.
Another helpful feature common in ecommerce, but rare in the enterprise, is the ability to personalize search results. The actual context available varies widely depending on your current environment, but the information commonly available within the firewall includes user name, job or role, location and even extensive internal search history. And, of course, search needs to respect access permissions across all enterprise content.
Enterprises provide ample access to user context. Most modern search platforms include the ability to incorporate that context, whether it be with proprietary tools or with machine learning tools like Apache Spark. A number of commercial and open source products include technology that will take advantage of that context.
How McDonald’s Drove Productivity Through an Elevated Employee Experience
In the new remote/hybrid workplace, work/life boundaries are blurred and workplace stress is a top driver of mental health needs.
How to Future-Proof Your Employee Experience Strategy in 2023
A framework to navigate through economic uncertainty
Challenges to Efficiency in 2023: Your Employees Need the Digital Workplace of the Future
The era of asking employees to do more with less is upon us
The Essential Role of Communicators in Fostering Wellbeing in the Digital Workplace
Join us for practical insights on how digital communicators can support employees to thrive in the digital workplace
Addressing Employee Needs and Wants with a Digital Workplace
The workplace is getting more and more digital – both in how we work and where we work
Maintaining a Human-Centered Approach During Digital Transformation
When it comes to digital transformation - people drive change, not technology
What context do you have access to that would improve search results?
One of the first sources of useful context for search in the enterprise — user metadata if you will — is about the person performing the search. Let’s take a quick look at the most common — and most useful — user attributes:
Job Title or Role
It makes sense that people who work in the same department with similar jobs will often be searching for similar — if not identical — content. By extension, employees in similar departments or roles will likely be searching for similar content. This is true for less urgent queries, such as the lunch menu, as well as for access security credentials that insure only employees with proper permissions see protected content. Performing an occasional search audit never hurts to insure access and permissions.
Co-Workers and Location
In the same way that colleagues with similar job titles and roles in an organization will often execute similar queries, departmental co-workers — or even co-workers with similar roles in other offices — will likely perform similar queries. So queries from co-workers or colleagues with similar roles are likely to search for similar content, so associating the commonly performed queries with search activity within the same cohort is often a winning option.
In addition to individual profiles, a user’s search history, previous queries, pages views and in the case of ecommerce sites, purchases, provide great information to make future search activity more relevant. If you can utilize even some of this history, you can provide even better relevance for your intranet search queries.
Remember Access Security
Finally, remember that regardless of how similar queries may be from a given department, different employees will often have different access permissions, so be certain that people have the proper access credentials for the results delivered.
Related Article: Diagnosing Enterprise Search Failures
How Do You Utilize the Metadata?
Businesses have access to a great deal of metadata to improve search results, but implementing it isn’t as easy as we’d like it to be. The good news is that tools like Apache Spark can be used to implement machine learning, essentially automating the task of improving relevance over time based on user activity. And while you may find you can prime Spark or other ML tools, these tools ‘learn’ relevance fairly quickly — in essence, taking advantage of user search behavior over time. Many commercial search products already bundle Apache Spark and its MLlib component, so you may already have the solution installed.
Learn how you can join our contributor community.
About the Author
Miles Kehoe is founder of New Idea Engineering, a vendor neutral consultancy focused on enterprise search, analytics and big data. In addition to New Idea Engineering, he has also worked at search vendors Verity, Fulcrum Technologies, and most recently at Lucidworks. Connect with Miles Kehoe: