Review: Funnel Back''s new search feature - Flusters

To provide a little background, Funnel Back is a search technology developed and commercialised by CSIRO.

It has been deployed in Australia.gov.au as their Whole-of-Government search technology as my agency's website search tool (as a hosted solution) and in many other agencies and companies across Australia and other countries.

It's a reasonably good search engine if some time is spent configuring it and I've been happy with the search success levels we achieve (though always trying to improve them).

AGIMO recently invited my agency to participate in the live pilot test of Funnel Back's new search feature - Fluster (50kb PDF).

In brief Fluster helps users find what they are looking for by offering alternative phrases to refine their search terms.

An example of this in action is visible in Australia.gov.au - simply use the search and look at the Related Search area at the right of the page.

We've been trialing this feature within our site for a little over a month now and I have an initial view on how Fluster has been performing.


How Fluster is doing
Initially I was concerned about the relevancy of the topics and phrases that Fluster would choose to display. This hasn't proven to be an issue, Fluster is providing highly relevant results.

However I'm not convinced that people are using the tool effectively. We've seen no measureable change in the search success rate and I do not have evidence that visitors to our site are using the Fluster Related Search area when searching.

This could be an education issue. We currently present Fluster in the search results page without any form of help, meaning that our visitors are not guided to the tool.

It could also reflect that improvements are necessary in the reporting of Fluster use so we can determine if the tool is assisting people find what they need. These reports are still being refined by Funnel Back.

Another factor I keep in mind is the trend towards more sophisticated internet users.

A large proportion of people are very familiar with Google and other 'generic' search engines and have learnt to use phrases rather than individual words to increase the relevance of results.

In fact, the average length of a search term in Google exceeded four words at the end of 2007 - at least according to WebProNews which reports that People Are Finding More Words To Search With.

This means that people are already refining their own search terms, potentially reducing the value in having a search engine do it for them.

In conclusion
So my preliminary conclusion is that Fluster can add value to search results.

However more time will be required to really understand the impact it is having and test ways to help people use it effectively.

While internet users are becoming more sophisticated, this doesn't negate the value of Fluster. There are always new people coming into the user pool and even experienced users may on occasion find that Fluster suggests a topic or phrase that they had not considered but leads them to a relevant result.