Minerva Studio - Fotolia


Can enterprise search tools achieve parity with Google?

Companies want an enterprise search function that performs like Google, but hurdles persist to make that scenario reality. Emerging technologies offer hope of future parity.

The ability to efficiently find and retrieve desired information is a cornerstone of enterprise content management (ECM), yet performance of enterprise search tools is often subpar. Why is that the case?

There are a couple of issues here. First, traditionally enterprise search tools have received short shrift in favor of technology that more directly delivers ROI, so the experience still leaves something to be desired in terms of speed and ease-of-use. Second, enterprise search often pales in comparison to top Web search engines—which isn't an entirely fair comparison, because they are distinctly different technologies with divergent purposes and practices in play behind the scenes.   

Be that as it may, enterprises are increasingly frustrated by this performance gap, and ECM vendors are taking note and trying to overcome some of the unique challenges that have hampered enterprise search in the past. 

The Google experience  

Gartner’s 2015 Enterprise Search Magic Quadrant recently evaluated 15 of the leading providers, and there are no real surprises with the usual lineup of vendors featured. Whether or not my clients are versed in the magic quadrant, their perspective will likely be the same heading into any search improvement project:  they'll want to bring in a Google-like search experience.

Before explaining why that scenario may not make sense for the enterprise, let's discuss the distinguishing characteristics of a Google-like search experience, which in most cases, it boils down to a few key ingredients:

  • Search accuracy. The perception that desired items will fall on to the first page of search results.
  • Recommendations/suggestions in search. The perception that relevant suggested searches or spell checks that relate to the original search query are made as they are typed.
  • Natural-language search queries. The perception that a full language approach of asking questions is possible and the search engine will respond in a meaningful way.
  • Sensible facets. The understanding that search results can be sliced and diced in a way that is meaningful to the underlying content and the original search query.

Let’s take a closer look at these ingredients in reverse order:

  • Facets. A facet is a dimension or classification related to a search result. Google’s first-level facets include Web, Images, Videos, News, Shopping, Books, Flights and Maps. The concept is simple – search results can be mapped by relation to Web pages, images, videos and so on. Second-level facets are contextual: date written and whether it's a blog item are facets of news; size and color as facets of image; price and seller as facets of shopping.
  • Natural-language search queries. A natural language search is the ability to enter a sentence (such as, “What is the average price of rental property in Chicago?”) and generate search results that are more accurate, based on the semantic meaning of the sentence rather than on keywords within the sentence.
  • Recommendations/suggestions in search. The continual suggestion of search terms and sentences as a search query is typed, either based on previous searches or by contextual meaning within the search query.
  • Search accuracy. While Google searches are fruitful, I usually need to play with the search terms to surface desired content in the first couple of pages of results. Often the top results fall into the category of information that people want me to find and those content owners have invested much effort in Search engine optimization (SEO) to facilitate their top ranking. Paid for advertisements will, of course, also occupy the first few search results

Challenges for the enterprise search engine

Since, as a consultant, I’m often asked to deliver a Google-like Enterprise search experience, I'd like to offer some perspective on the differences between Web and enterprise search tools:

  • Facets. Google facets are reasonably useful for Web content, but less useful for enterprise content. In the Enterprise, facets need to be directly related to the business context of content, and enterprise content is largely documents-- not Web pages. In the enterprise, facets need to be determined through a taxonomy that provides categories for the facets engine to surface. Complicating matters is that a successful facet for one business may differ entirely for another. A property management company that manages properties, tenants and leases, for example, may have those content categories as facets. A medical equipment manufacturer that manages equipment models, assemblies and parts-to-customers will likely have those categories as facets.

The key is that a search engine cannot figure these facets out – those designations needs to be guided and augmented by a taxonomy that is developed by people

  • Search accuracy. Effective SEO techniques, often the work of content producers, improve search accuracy. For a Google search to return results there are two ingredients – the originating content needs to have been tagged accurately using meta tags – these are typically keywords, title and description of content at the Web page level-- and the Google calculated page rank needs to be of a relevant value. On the Web, content producers will make extensive use of meta tagging – it's their job to make sure it's found. The page rank implies an authoritative source of content, through linking to reliable sources of information – this is the foundation of the Google search paradigm.

In the enterprise, without a search strategy and a recognized taxonomy, content producers will not tag content. It may not even occur to content producers that their content should be findable. Also, most enterprise content is likely unstructured documents, stored in line-of-business applications such as ERP and CRM, not structured Web pages and data. In this context, the concept of page rank breaks down – there simply isn’t enough cross-linking of content to facilitate effective page rank analysis.

  • Recommendations/suggestions in search. Google is used by millions of people daily and this provides its search development team with massive insight into search behavior – what people are searching for and how they searching for it. In the enterprise, examining and analyzing search logs is a critical aspect to an effective search strategy, but the reality is that most companies don’t have large search development teams reviewing search logs;  Search is Google’s business – it is not the business of most enterprises.

 Jonathan Bordoli offers further perspectives on providing a Google-like search experience in the second and final part of this series

Next Steps

The hurdles facing enterprise search improvement

Content analytics should improve enterprise search

A roadmap for building an enterprise taxonomy

Dig Deeper on Enterprise search platforms