Hello everyone! I am an AppFoundry partner and I am currently building a tool that helps supervisors monitor the effectiveness of their Agent Assist knowledge base. We want to track how often agents are actually clicking on the suggested articles and if those articles are helping them resolve the customer’s issue. Is there an API endpoint that provides ‘Click-Through Rate’ (CTR) data for Agent Assist suggestions, or do we have to build a custom event listener in the agent desktop?
Hey Koj59! I am a troubleshooter and I spend a lot of time with the agent desktop. There is no native API that gives you a pre-calculated CTR for Agent Assist. You have to listen for the v2.users.{id}.agentassist.suggestions events via the Notification API. When an agent clicks an article, the desktop sends a feedback event that you can capture. You will then need to correlate these clicks with your interaction data in your own database to calculate the effectiveness.
I have designed several of these feedback loops. To follow up on Val32, you should also look at the ‘Article Relevance’ scores. If the AI is suggesting articles with a low confidence score and the agents are still clicking them, it might mean your knowledge base has some gaps that need to be filled. We use a custom dashboard to show the supervisors which articles are being ‘Ignored’ vs ‘Accepted’ so they can prioritize their content updates.
Greetings. I am a migration consultant and I have seen many organizations ignore this feedback data. Koj59, if you are building a tool, please make sure it also tracks the ‘Resolution Rate’ for interactions where an article was clicked. If an agent clicks an article but the call still ends up being escalated, the article itself might be outdated or incorrect. This is the ‘Holy Grail’ of knowledge management analytics!