Skip to main content
valentina_montoya
Fronteer
August 10, 2026

Using tags and your knowledge base to cut down on repeat questions

  • August 10, 2026
  • 1 reply
  • 12 views

One workflow that works really well for teams drowning in the same handful of questions: closing the loop between your tags, your analytics, and your knowledge base.

The idea is simple: Set up rules to auto tag inbound conversations by topic (billing, onboarding, integration questions, whatever your common buckets are). Then check those tags in analytics every month to see which topics actually drive the most volume. That tells you exactly what to document.

From there, create two types of articles for the top topics using your knowledge base. Internal ones so every teammate answers the same question the same way, which makes responses faster and more consistent across the team. And customer facing help center articles for the same topics, so customers can find the answer themselves before they ever reach out.


That last part is where the repeat question volume actually drops. The internal articles make your team faster, but the customer facing ones are what stops the question from landing in your inbox in the first place. Over time, your most common topics become self-serve, and your team spends less time on questions that don't need a human to answer them.

Easiest way in is to start with just two or three tags so it stays manageable, then expand from there once you see the pattern.

1 reply

ejayr
August 11, 2026

Hi ​@valentina_montoya,

Solid loop, and it holds up. Worth adding for anyone copying it: Topics can handle the auto tagging step now, since Front AI labels conversations by subject and rules can fire when a topic is detected.

On the two article types, the split matters in practice. Internal articles work with the sidebar plugin, but the composer can only insert from a public knowledge base that is published.

For choosing what to write next, the Knowledge base report on Professional and up lists the search keywords people actually type, which is a faster signal than tag volume alone.