You can pass the cool.X parameters interactively via CGI when you run a query, so this can be a good way to test how different weightings will affect the result set.
Generally speaking tuning the ranking algorithm is the best way to get the best overall improvement in your result set because there is a much more rigourous process involved when determining the optimal ranking settings.
However if you do want to adjust the values manually adding the cool.X parameters will show you immediately how the results are affected.
Thanks for that - I’m struggling to see much change using the cool.X parameters, so perhaps the problem wasn’t really much of a problem. I think I’ll take your advice and start learning more about the tuning - seems like a more gradual and understandable way to work with our results. Hopefully I can get more people our end involved in that.
Generally speaking - before you look at the ranking algorithm it is well worth looking at a few other factors which can have a far greater impact on ranking.
If you have control of your content then consider adding noindex (html comment) tags to the template so that you hide headers/footers/navigation from Funnelback. Removing the irrelevant content will have a big impact on what is identified as a relevant result (eg. a search for contact won’t potentially return every page because it’s in your navigation). PaDRE (Parallel Document Retrieval Engine) - Funnelback Documentation - Version 15.12.0 explains how the noindex tags work. This technique removes irrelevant content (noise) from the index making the results much more focussed.
Use synonyms in conjunction with your analytics to feedback and improve the search. See: https://www.funnelback.com/blog/5-simple-tips-to-make-the-most-out-of-your-search-analytics This blog post I wrote a couple of years ago covers some of the basic techniques. This technique bridges the gap between user and organisational language by mapping user words to the words you use on your site meaning that the user’s search terms match the words in the index.