Skip to main content

How to use Tagging Guardian feature

Written by Vijay Krishna Mallik

The Tagging Guardian is a new Copilot capability that helps editors apply the correct advertising and editorial tags with a single click.

By automatically suggesting standardized tags based on a story's content, this feature reduces manual work, improves tagging consistency, and helps ensure advertising campaigns can reliably identify tentpole content.


The Problem

Editors occasionally apply incorrect or inconsistent tags. For example, using "NY-Fashion-Week" instead of the standardized "NYFW" tag.

Because these errors are often difficult to detect, they can prevent advertising campaigns and downstream systems from finding the intended content. The results:

  • Missed advertising opportunities

  • Analytics performance metrics were using multiple tags pointing to the same tentpole event

  • Inconsistent content organization

  • Additional manual cleanup for Editorial, Product, and Advertising teams


Goal

  • Make it effortless for editors to apply the correct standardized tags without disrupting their existing workflow.

  • With one click, editors can populate the appropriate advertising and editorial tags directly within Copilot.


What's New

A new Autofill Tags button has been added to the Tag Field section in Copilot for articles, July 2026. When selected, AI analyzes the draft and automatically recommends the appropriate tags.

The feature can apply:

  • Commercially approved advertising tag

  • 🟣 Suggested editorial/content tags

  • Editors can still manually add any additional taxonomy tags that are appropriate for their story. They will be marked in black.

All tags remain editable and removable.

Recommendation: Editors should avoid removing ⭐ starred tags, as these are required for advertising targeting and campaign delivery.


Tagging Logic

  • Story headline contains a recognized tentpole name

  • Story body references a recognized tentpole

Existing Logic

  • Story body contains keywords already recognized by our existing content classification models

Did this answer your question?