We started to include intranet search – for people and content – as part of the standard IBF design evaluation in 2006. Now in August 2007, we have seen quite a few different implementations, but with only a few exceptions, intranet search is still very much a poor cousin to the public internet variety. People search, in particular, is fond of declaring ‘no results found’ when a user has made only a simple single-character typing error, leaving them with no choice but to try various permutations until they get lucky. At least with a printed staff directory, frustrated users could thumb through until they found what they needed, but most intranets lack the electronic version of this fallback solution. Quite simply, if you don’t know how to spell the name you are looking for, you’re stuck.
Ironically, one of the solutions we have been persuading intranet managers to take up is almost a century older than intranets themselves. Soundex is a fairly simple set of rules to turn similar-sounding names into a four-letter code. It does have some drawbacks and should not replace a simple text match as the first step in a search, but it is implemented by most scripting languages and is fairly forgiving of spelling and typographical errors.
Content search does a little better in our scoring, but even here we find poor regard for the user experience, with no mention of the original text searched for and scolding messages for choosing search terms that yield too many results. Yet public internet search is very big business indeed, with companies like Google and Microsoft falling over each other to retain or capture market share, respectively. The driving force behind this enthusiasm is advertising. Unfortunately, there are no similar forces at work on intranets. While a few enterprising organizations do actually carry third-party advertising on some pages, the economic model for getting search right is completely different. It is all to do with saving time and frustration within a captive audience. So what is needed is a means of measuring and improving the perceived value of search within an organisation. Knowing how long a user has spent searching and whether they were successful would be a step in the right direction, but there is little evidence that even this basic level of measurement is taking place. Perhaps this is something we should be including in our benchmarks? Let us know what you think.
