Hello everyone, I'd like to know what is the best way to optimize the recommendation engine of Responsive so that it could give us the best possible answer. Is there anything we can do/change about it or is it a fully automated feature?
@Juan Reinholdsson we found it best to structure our response headers in the library AND in the imported file. For example….we were at first using the prospects terms on imported items…comments…response…explanation etc. now we often change the imported file title to "comments" for example…and all our narrative content in the Q&A library using "comments" as well…so they align. We also run auto respond by sections…and adjust the tags per section and that has helped. @Sarah Biskup anything you would add?
@Juan Reinholdsson We have a couple resources in the Academy you may want to check out:
Coffee Chat on the Auto Respond Feature (35m) -discusses how search is utilized
Optimizing Auto Respond (5m)
Also - not sure if you have ever checked out the Content Score - which allows you to customize what the search engine should prioritize
Finally, I'd also recommend and updated/clean library - you can review some tips here!
I agree with @Andrew Mersman especially with auto respond. As far as just the recommendations pane goes, we did a pretty big tag overhaul that improved our recommendations greatly. We had a lot of generic tags and we really restructured - for example, we had a Corporate tag that contained general corporate content for all solutions - but we broke it out and have a Corporate - Post Acute and Corporate - Human Services as well and we select either of those 2 tags depending on the project. We also generally only have 1 tag per Q&A pair which has helped. It was a lot of work up front but there's a huge payoff with recommendations as well as auto respond.
We also utilize the content score like Helene mentioned above. We go in quarterly and update the star rating on the Q&A pairs based on the content score. We set in Org Paramters how we want the system to generate our content scores.
I'll update if I think of anything else! Hope this helps.
