Skip to main content

​
Direct answer

Use a Topic as the primary semantic grouping for a prompt. Use Tags as reusable labels that can cut across topics, such as audience, product line, or campaign. Use the prompt’s funnel stage for buyer-journey analysis. Design these fields around decisions you expect to compare, then keep the names stable.

​
Data required

  • The current prompt inventory
  • The product areas and use cases used by your team
  • The audience, market, and campaign dimensions you need to filter
  • The funnel-stage convention your team will apply
  • An owner for taxonomy changes and a list of accepted names

​
Workflow

  1. Draft a short topic list where each topic represents one durable business area.
  2. In Workspace > Topics, create or consolidate those primary groupings.
  3. Define tags only for dimensions that need to span topics. Manage them in Workspace > Tags.
  4. In Workspace > Prompts, assign one primary topic, the relevant reusable tags, and a funnel stage to each prompt.
  5. Open Cockpit > GEO Matrix and Cockpit > Compare. Apply each topic, tag, and funnel filter to confirm that the resulting groups answer a real question.
  6. Review unassigned prompts and near-duplicate labels before using the taxonomy in exports or Looker Studio.

​
Interpretation

A useful taxonomy produces groups that are distinct enough to compare and large enough to inspect in context. A topic with no answered prompts is an empty state, not poor performance. A group with very few answers can change sharply when one answer changes. When comparing groups, preserve the same period, providers, models, countries, and competitor classification. Product Mention Rate and Citation Rate are answer-level metrics with response-based denominators; Product Coverage is prompt-level. Do not treat those percentages as interchangeable.

​
Possible next actions

  • Test whether merging duplicate tags makes filtered reports easier to interpret.
  • Test whether a broad topic should be split when it contains separate decisions with different answer patterns.
  • Add a reusable audience tag when the same segment appears across several topics.
  • Document a naming rule before allowing multiple editors to extend the taxonomy.
Treat taxonomy changes as hypotheses about better analysis. Record renames and merges because they can change historical groupings.

​
Limitations

  • Topic and tag assignments describe your taxonomy; they do not prove market demand or business importance.
  • Funnel-stage labels require consistent human judgment.
  • Small or uneven groups can produce unstable comparisons.
  • Renaming or reassigning prompts can make a trend look different without any underlying answer change.