Debunking common enterprise search myths

Guest post by Ted Carroll
Search offers the opportunity for delivering significant benefits to users, but is usually a neglected aspect of ECM solutions. Despite the huge amount of interest in search engine optimization (SEO), there is limited interest in the equivalent tailoring of “corporate search” results--for an Internet site, Intranets or internal document management / content management solutions. The sheer number of users on well-used Internet and Intranet sites should make such optimization potentially interesting, either through reducing internal staff costs, reducing alternative channel costs (e.g. phone, email, post) or generating more business from customers.

Most Internet sites do have a search capability and all ECM/EDRM solutions have search built in, but for what benefit? The low cost of providing a search capability makes it all too easy to ignore the value of information being served through search. However, each page on a site has significant content creation costs associated with it, which is wasted if the page or document is not found by those users attempting to locate it through a search function.

Search is often seen as simply “technology” that is plugged onto an existing solution and, once in place, there is nothing that can be done to influence and improve the quality of results. Nothing could be further from the truth. We can influence search results and we can tune a search engine to deliver better results, which will deliver additional benefits to the business.

Let’s look at the myths

How have we gotten to the point where search is so neglected ? A number of views could be responsible:

“Search is a black box” -  The Google web search is a black box as far as most people are concerned-–through the use of PageRank-based relevancy (which takes little or no account of the content on the page), it is difficult to influence results by changing content. However, with any good “internal search” engine (including those from Google) , it is possible to improve the quality of search results in a number of ways including:

• Improving content quality, in general (including metadata);
• Providing high quality titles and summary snippets; and
• Tuning the search engine i.e,. the approach used to weight title, summary and metadata (or change the search engine).

“Search terms are too varied” -  An analysis of search terms will show that this is not usually the case. The classic “long tail” is typically what search analysis shows-–a smaller number of common search terms followed by a long tail of less frequent terms. Clearly, it makes sense to concentrate on solving problems with the common search terms first, as this is where the most benefit will be delivered.

“Language is too complex to understand search needs” -  Language is infinitely complex and there are, therefore, many ways in which a search could be constructed. However, a group of users will generally use similar search terms when looking for the same information (as identified in the point above.) Search is essentially a text-based matching process--a search engine is looking for strings of text that match what the user has typed in--search engines (mostly) do not carry out sophisticated semantic analysis of either the search term being entered or the results. Understanding search terms and improving the content provided is a relatively simple problem that can be analyzed and improved.

“Search does not deliver benefits” - Search applications have been delivering benefits for many years, but in most cases, the benefits are not measured. One reason is that search solutions have limited capability to log search usage, particularly when embedded in applications such as EDRM or collaboration sites. Vendors need to do more to help sites measure search usage, with “out of the box” reports to summarize common search behaviors. A recent piece of Forrester Research (“ECM Priorities for 2010”) claimed that  “49 percent of organizations couldn't estimate the ROI for any of their ECM systems, making it difficult to get money for expansion.” Clearly system owners should do more to articulate the benefits from ECM solutions, and that includes search benefits.

“Search is not important” - Search is an essential component for the World Wide Web and it is difficult to see how the Internet would work without comprehensive search engines that deliver meaningful results. Why then, in the corporate world, is so little effort spent on delivering search and the equivalent of search engine optimization? Increasingly, search is an access model that users understand through the use of Google and other search tools used when at home. Search provides an essential component of the “mobile” Internet world, where growing personal use is being mirrored through corporate use of mobile devices that move beyond email to application access. A further interest for search is that voice-based interfaces will rely increasingly on search as the method of access. How do you navigate a file plan using voice: “Down a bit...left a bit...”?)

“Search results are always poor” - The truth is that many ECM search tools in the past have been very basic and delivered poor results (partly as a result of the technology, but also due to poor content management practices.) This is a shame, given that search applications were the foundation of the document management market (tools such as Basisplus, BRS/Search, Assassin, Status, Stairs) before the emergence of File Plan based applications such as Documentum, Saros (now FileNet), PCDocs (now OpenText). Search tools have made gains recently, but there is still much room for improvement, and content management practices that can improve search results are still lacking in many cases.

“Search is a moving target” - Search results change depending on what content is available on a site at a point in time. Any site with significant content is changing on a regular basis--the search results for a set of terms may be perfectly tuned one day--until another department adds related content and upsets the ranking of existing content. This is a big challenge and one the industry needs to address through better search analysis tools.

“Search cannot be measured” -  How do you measure and assess search? According to some accounts, Google uses armies of test users, but clearly, few organizations can invest such resources. This is one of the more challenging areas. It is not easy to deliver a method of measuring the quality of search results. I think we will see solutions emerging in this field in the next few years.  

Moving forward

I hope that, as you have gone through the above list, it has gotten you thinking that maybe search is worth more thought and effort than at present. But how much do you know about the users searching your applications? Can you answer the following questions in relation to your content management applications?

• Do you know the total number of searches carried out each day/week/month?
• Do you know the proportion of users that have at some time used the search facility on your site or application?
• Do you know the proportion of questions (or information requests) that are answered using search, as opposed to other methods (e.g. using page/file plan navigation)?
• What are the top 50 search terms used?
• What proportion of searches produce no results?

If the answer to these questions above is "no" or "I don't know," but you have a bunch of very satisfied users or customers, then maybe there is not a problem. However, if you have less that satisfied users, it may be worth delving into search activity and understanding how search contributes to the overall user experience. You may be surprised!

Ted Carroll has an extensive background in content management working for software companies, systems integrators and as a management consultant in the United Kingdom. He authors a search related blog at www.tedcarroll.org and has developed new approaches to the measurement and assessment of search.

Related Articles:
PerfectSearch debuts with fast disk-based federated search
Coveo 6.1 expands search into Enterprise 2.0 data
Traction teams with Attivio for secure social search
Nstein launches semantic search tool
Coveo releases free enterprise search tool
New Bing feature helps you visualize your search