We had some development done on our site in order to specifically by pass the funnelback “algorithm”, this is within the metadata of that page and uses the 1-10 scale. I did some work yesterday on sorting the order the products appear in the View All Products | Temporary Works Equipment for Hire section using ‘Product Upweighting’ in Metadata.
I assigned the numbers as attached but they aren’t all working. The idea was for the groundworks overview to be first, followed by category overviews and then products (which I also assigned numbers to so that groups of products would show together rather than randomly).
It looks like some of the product groupings have worked, but not all. And it is displaying the existing category overview page (trench sheeting and light piling) after some of the products.
Funnelback’s usual upweighting mechanism is reliant on the keyword matching content within a field. So if you apply an upweight to a metadata field then the upweight will only trigger if the search keyword appears in the content of the field.
It looks like what you are trying to do would be achievable by creating what’s called query independent evidence, where a weighting is assigned to individual URLs or sections then this provides an upweight or downweight regardless of the query terms.
Further information on QIE is available here:
This KB article might also provide some useful hints on things you can do:
For query independent evidence to be applied you would need to extract out the metadata values that you have embedded in your pages and produce a qie.cfg file within Funnelback.
There are a few approaches you could take:
create a qie.cfg file on your collection then add the URLs and individual weighting to the file.
If the metadata is in a CMS, you might be able to produce the qie.cfg file in your CMS dynamically. If so, this can be made accessible at a specific location and FB can be configured to download the file using workflow as part of the update.
this is a bit more involved, but it’s possible to use the metadata included in the FB index to generate the qie.cfg file as part of the update workflow. In a nutshell you create a custom template in FB that outputs a qie.cfg file when you run a search. You then write some workflow to run a search that returns everything, and run this against the offline index after the new index is built, saving the output as the qie.cfg file. Workflow then needs to apply this qie to the newly created index. There is a bit of a tutorial in the knowledgebase for a similar scenario, where the FB index is used to produce auto-completion CSV. Redirect Notice The concepts in here are similar (though a little more complicated than what you would need) to what you would need to do to produce the qie.cfg.
If you don’t have backend access to your Funnelback server, or don’t feel that you’re up to doing this then get in touch with your account manager and I’m sure someone within Funnelback would be able to assist you with writing the template and workflow.
If you’re looking to use Funnelback’s system-generated queries behaviour, then what you’re describing may more closely match a custom sort order, rather than tailored ranking. If a qie.cfg file already exists, it may be able to serve a dual purpose (custom sort and custom ranking).
Assuming your initial system query is something like: productCategory:shoes , there may be very specific products or sub-categories you’re wanting to prioritise here.
External metadata for a custom sort may be appropriate, with entries like:
If there’s already a custom workflow in place to retrieve the qie.cfg file, it may be easy enough to extend this workflow to also output external_metadata.cfg.