Skip to main content

Hey there!
I noticed that the order of search results in my community seems to follow an incomprehensible pattern.

When I search for a term the results neither seem to follow the order of most recent, most relevant/ viewed/commented nor most accurate contribution. There also is no pattern regarding the role of the user, who posted it.

Therefore it sometimes shows an old post, that’s only viewed by few people and commented by none on the very top, a highly viewed and commented post on the third position and a newest contribution on the last position.

While I would understand the logic of the newest post at the very bottom, if it was about the exposure of the post, this doesn’t line up with the rest of the order.

So, does anyone know, what’s the actual logic behind the order of search results (community posts as well as Zendesk articles)?

Best,

I have noticed this, too.


Good question.

If I’m not mistaken, search in Gainsight CC is powered by Algolia (they whitelabel it). Maybe there’s some docs on it that can help clarify. 


Hi @Xentral_Community - as already mentioned by @DannyPancratz we are using Algolia to power our community search.  Algolia utilises an algorithm that considers and weighs-up numerous factors before presenting the search results, which is why the ‘pattern’ is not so easy to discern.  For example, there are a number of searchable attributes that are used to index topics such as the topic title, topic content, replies, tags, category, etc…  The algorithm will then take a number of criteria into account before deciding on the search results order.  For example, the number of words from the search terms that match, the positioning of the words, likely typos, when the topic was last updated, the number of views and replies, etc…  Questions with a marked best answer and articles will be given some priority in the results, as they are most likely to have a high quality answer.  All of the complexity that’s needed under the hood to deliver search results can mean that in some cases we see a result that on the surface doesn’t make sense or isn’t the topic that we’d want to be surfaced.  In those moments, a little bit of curation can help a lot.  For example, adjusting the topic titles of important topics to more closely match the search terms visitors are using, or archiving content that is old or irrelevant and is cluttering up search results.  


Reply