Dynamic Reranking

Dynamic reranking uses machine learning to re-rank results and answers based on user engagement data. Our team has trained a complex model that reranks FAQs in order of their likelihood to be clicked by the user. The best part? This model will become smarter over time as it is fed more user interaction data.

For example, if it sees that the most clicked on entity is the one that we’re returning in the fifth position, this new feature will allow the algorithms to dynamically, re-rank the results and bump that result to the top because the goal is always to have the first result be the one that’s most relevant and most engaged with.

To activate this feature, you can add a new object called dynamicRerank in your Search Configuration file with a target object like so:

 “faqs”: {
“dynamicRerank”: {
“target”: “CLICKS”

This means that it is optimizing for the most clicks, but in the future, you will be able to target other events like portray a particular conversion action or a type of CTA click.

*Note: This will only apply to FAQ entities in the Spring ‘21 Release, but stay tuned, in a future release, the reranking model will be able to rerank more broadly across all entity types!

Have any feedback or questions? Leave us a note below!

1 Like