Some Searches Are More Equal Than Others

I was perusing my boss's blog the other day, and I came across a review he wrote of a presentation by Google, a company everyone has their eye on (especially if you work in the Microsoft "ecosystem" like we do).  Google is attempting to make inroads into Enterprise Applications (Search, Business Productivity, etc.), and David was learning about their strategy for doing so.

One point I found interesting in David's recounting of the presentation was that Google is selling their enterprise search tools by selling the knowledge that they have gained from internet search.  I found worth drilling down into a bit.  Just last week, I had a discussion with some reps from the search vendor Coveo about this at the Gilbane conference in Boston about how dramatically enterprise Search and internet Search differ. It seems that they sell their enterprise search tools in exactly the opposite way.

I did a little research, and (of course!) found a white paper [note: it's a link to a document] from Microsoft that helps underscore some major differences between enterprise search and internet search, namely:

• Link structure.  Content on the Internet is much more likely to be linked to other content than it is within the enterprise, where data resides in documents or line-of-business systems not typically linked to other documents or data sources.
• Site structure and hierarchy.  Content on a corporate portal is much more likely to be stored in a repository that is more narrow and hierarchical, whereas on the internet the hierarchy is much broader and flatter.
• Security.  Much content published to the internet is available to all.  This is virtually never the case within an enterprise.  Security in the enterprise is a top priority, in direct opposition to the "open" nature of the internet at large.  Nearly every engagement where KMA has worked with customers to deploy search as part of SharePoint (or separately), the customer has learned unpleasant things about their existing security infrastructure in the process -- the projects very quickly surfaced issues that required corrective action.
• Relevance.  More generic measures (e.g., high weighting to link structures) may be applicable for calculating relevance in large-scale, consumer-facing internet search, but enterprise search requires relevance that is highly configurable and can be tuned over time to match the patterns and quirks of people who search for information within an organization.
• People.  As often as not, an enterprise user seeking knowledge about a particular project, product or process seeks a subject matter expert for a dynamic exchange of ideas or Q&A about a particular topic, rather than a link to a static document, web page, etc.  Being able to find these subject matter experts is a critical requirement for effective enterprise search.

As a KM firm, we've built search engines, helped our customers assess and compare search solutions, and designed and deployed solutions incorporating search.  As Microsoft's search solutions mature, I look forward to learning more from our customers about unique use cases driving search needs. We've invested a lot of time and money in training people on search tools, but it's our listening to how users need to find enterprise data helping them do so that will make us most valuable to our customers.