Current Style: Standard
Ten enterprise search implementation success factors
Compiled by: Martin White martin.white@intranetfocus.com, December 2011
1. Establish the search support team
Even a low cost search appliance will need at least 0.5FTE support, and a high-end enterprise search application will probably require a team of four. If these resources are not available there is no point in making an investment that will never produce good searches time after time.
2. Research user requirements
Some users will need to find a particular document, and others may like to see a range of results to help them decide what they really need. Understanding user requirements is essential in defining the technical requirements of the search application and any customisation that will be required for the user interface.
3. Don’t buy more search capability than is actually needed
Functionality that is not needed is money wasted and complexity enhanced. Most current search engines provide more than enough capability without needing to buy a high-end enterprise application.
4. Understand the technology of search
Every search application makes use of computational linguistics, applied mathematics, and software and hardware engineering in subtly different ways. Understanding how can help in making the best use of these technologies to achieve user satisfaction with the results of a search.
5. Plan carefully for secure search
Searching for [confidential] in a search application might well identify information that should indeed be confidential. Making sure that confidential information remains confidential is one of the most challenging aspects of search implementation.
6. Search is a dialogue
Rarely will a user find what they are looking for in the first few hits when using a single search term. Search effectiveness is achieved through enabling the user to maintain a dialogue with the search application through a carefully designed user interface that meets specific requirements.
7. Best bets are not the total solution to search success
Although providing ‘best bets’ at the top of a search list can be helpful to users the moment they trust a best bet and find that it is not, then all best bets will be viewed with scepticism. Best bets need to be reviewed and revised on a regular basis in conjunction with subject experts from within the organisation.
8. Search analytics need to be reviewed and actioned
Reviewing search logs to identify, for example, the top 50 search terms, and searches that resulted in no downloads is time consuming but essential to maintain a high quality search service.
9. Federated search presents significant search challenges
Successful searching across multiple repositories, and especially multiple applications, is very difficult to implement in a way that produces consistently relevant results.
10. Monitor changes in user requirements
As business priorities change so will the requirements of users. These need to be identified and actioned as quickly as possible so that changes to relevance algorithms can be undertaken.
Resources
Martin White. Making Search Work. Facet Publishing Ltd, London, 2007. http://www.facetpublishing.co.uk
Martin White and Stephen Arnold. Successful Enterprise Search Management. Galatea Publishing. Manchester, 2009. http://www.galatea.co.uk
Louis Rosenfeld. Search Analytics for your Site. Rosenfeld Media, New York, 2011. http://www.rosenfeldmedia.com
Rocardo Baeza-Yates and Berthier Ribeiro-Neto. Modern Information Retrieval: The Concepts and Technology behind Search. 2nd edition. 2011. Addison Wesley.
Peter Morville. Search Patterns. O’Reilly Publishing, San Francisco, 2010. http://www.oreilly.com
The Real Story Group (formerly CMS Watch) http://www.realstorygroup.com
Pdf Version:
