concentration with headphones on

Your Search Platform is Talking to You: Are You Listening?

4 minute read
Miles Kehoe avatar

Quick: What's the last search query submitted to your search platform?

If you manage a public-facing company website, I bet you can answer this question in seconds. And I’d also bet you could tell me how many times that term was used in a query in the last hour, day and week, who did the search, and maybe even what documents the person viewed after the search.

Public-facing search is often seen as generating revenue or good will for the organization. Armies of companies exist that will — in some cases for thousands of dollars a month — use their special tools to give you all of the metrics and reports that management loves. 

If only internal-facing enterprise search could make money for the company!

What's the ROI of Preventing Litigation?

The bad news for enterprise search, including intranet and extranet sites, is that most of these tools are just not available. Open source search technologies like Lucene and Solr have replaced the old commercial standards, Endeca and FAST. But replacing a product with a handful of programs doesn’t provide the same information.

So what do e-commerce companies, where search analytics are critical, do when they adopt open source platforms? Generally, they write their own reporting tools. And even though you may object, e-commerce and e-discovery generates revenue and helps avoid litigation. Few companies recognize the hard ROI of not being able to find a policy or a partner agreement.

Can you justify spending money to understand your intranet search activity? I’d propose that there are risks for not knowing what your internal site users want to find.

Imagine a scenario where a problem employee in a remote office is harassing other employees. Staff in the remote office searches your intranet for harassment policy, but can’t find it. I’m not a lawyer, but as I understand it, if a company should have been aware of a problem, it can be liable for damages. One guarantee is that when this sort of thing happens, an organization will quickly find budget for ongoing analytics to avoid future liability.

When I mentioned analytics at one of my workshops a few years back, an attendee from a large US manufacturer spoke up. A few years earlier, the manufacturer had experienced a huge PR nightmare because of a product defect that led to millions (billions?) of dollars of penalties and fines. Just prior to my workshop, the company in question started a search activity log analysis and tested it retroactively over years of data. The attendee said the search logs showed a pattern of the defect 18 months before the story broke in the press and became a big, expensive and embarrassing legal matter. 

Learning Opportunities

Would that have been worth a few bucks up front?

You can find a number of other examples to justify investing in search analytics. Repeated or abandoned searches on your intranet or extranet sites mean that your employees and partners cannot find the information they need in a timely manner. What’s the cost of a tech support rep doing three or four different queries to "game" the search platform into finding content the rep knows is there? A partner company comes into your extranet and cannot find the policies for their customer to use your product — the sale takes longer. What is that cost?

What’s the Answer?

Many of the most popular search platforms in use in corporations today don’t provide the reporting tools that traditional e-commerce platforms offer. But don't accept failure and give up, or transfer to your public-facing web group. Start with a few simple reports that can be gathered from existing log files.

Your existing system may have some lightweight search reports, like the most frequent queries and the number of results. But even if your search platform doesn't include any reporting, there are almost certainly log files you can access that — when properly analyzed — can get you the bare essentials including:

  • Most popular queries
  • Queries that produced no hits
  • Frequent search users
  • Documents viewed following search
  • Abandoned searches

Some of this information is available in raw form in your server, CMS or search application log files. Break out your old Perl book and write a simple script to "glue" the search log results together with your server logs every few minutes, and offer a handful of reports. 

That alone will give you what you need to start improving your search results and reducing failed searches. Produce reports to improve management decisions. If you get really energetic, use one of the open source tools like Kibana or Banana, index the logs, and make some interactive graphics reports. If you're lucky, one day you may even get funding to expand the internal search analytics.

Creative Commons Creative Commons Attribution 2.0 Generic License Title image by  garryknight 

About the author

Miles Kehoe

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 searchvendors Verity, Fulcrum Technologies, and most recently at Lucidworks.