Partner Article

Enterprise search implementation: success factors

Iain Fletcher, VP of marketing for Search Technologies, shares his expertise on enterprise search.

The previous article reviewed the findings of major enterprise search surveys over the last 10 years. It concluded that good ROI and business productivity gains could be achieved from enterprise search, but only if the implementation was effective. The following article will examine some of the key drivers of effectiveness.

Dealing with Diversity

Within a typical enterprise, data is highly diverse and heterogenous. Data sets range from tightly-controlled customer databases to relatively lawless file shares and from records just a few words long to technical documents which may run to hundreds of pages. User diversity is similarly wide, from fact-checking administrators to full-time researchers.

So how can we ensure that such a wide range search user needs over diverse data sets will produce
great ‘findability’?

Top performers with enterprise search – those companies who derive substantial on-going benefits
– share a number of common approaches:

• They understand the need to customise the system to their environment, in order to take account of specific data sets and user needs. Every company is unique in this respect
• They accept that enterprise search, especially in a large organisation, is a potentially complex task that won’t work optimally out of the box
• They understand that there is a constant churn in data sets and user needs, so they ensure they are in a position to easily adapt
• They know that the bigger data sets become, the harder it is to deliver effective search

Out of the Box

Companies often go wrong with enterprise search because they put too much faith in the out of the box capabilities of the search engine. Vendors are keen to suggest that their products can just plug- and-play, and buyers understandably have a strong preference for this.

For a range of search applications - especially those which search a relatively homogenous data set and serve a group of users with a focused and definable objective - out of the box can perform perfectly adequately.

However, true enterprise search demands more attention to detail.

Customising Enterprise Search

Today’s leading search products are highly capable and provide a wide range of customisation options. This article will not review these in detail: however, there are a few issues that a best practice enterprise search system will typically address:

• Capturing and/or creating a lot of accurate metadata to drive search navigation and results sorting options
• Providing different user communities with customised relevancy ranking to suit their needs and save time
• Monitoring system usage - for example, logs of the most popular search queries or queries which fail to find any documents - and tuning the systems over time with continuous, small improvements
• Ensuring that document level security is fully respected and its implementation well- thought-out

For those interested, here is a detailed review of relevancy ranking factors.

The underlying technology which supports the points above is mature and stable, and best practices for implementation and tuning have been established. The key mistake made by most organisations who fail to achieve good ROI, is failing to recognise the need to customise enterprise search to their environment.

The importance of these implementation factors will increase with the growing amount of corporate data that search systems need to cope with.

The final article in this series will look to the future and consider how the enterprise search world is becoming entangled with “Big Data” applications.

This was posted in Bdaily's Members' News section by Iain Fletcher .