Browse documentation

Prompt Labels

Labels are tags on your prompts that let you ask questions like "what's my visibility on branded queries vs generic ones?" or "how do I perform on early-funnel discovery prompts vs purchase-ready ones?"

Your prompts aren't all the same — some ask about your brand directly, others are category-level questions where you hope to appear. Labels make this structure explicit so every chart and metric can be sliced by the dimensions that matter to your strategy.

What Labels Are

Labels are flat, cross-cutting tags that hang directly off prompts. They are NOT a hierarchy layer — your data model stays Tenant → Asset → Topic → Prompt. Think of labels as metadata dimensions for filtering and grouping, similar to tags on a blog post.

Each prompt carries exactly one value per label dimension. Labels are auto-derived when prompts are created but can be manually overridden at any time.

Branded vs Generic

The Branded label answers: does this prompt mention your brand by name?

  • Branded: "Is Allbirds Tree Dasher good for running?" — directly asks about you
  • Generic: "Best sustainable running shoes 2026" — category query where you compete for inclusion

This split is critical. Branded visibility should be near 100% (it's your name). Generic visibility is where GEO strategy earns its keep.

Intent Taxonomy

The Intent label maps each prompt to a stage in the buyer journey. A2Z Reach uses an 11-layer taxonomy derived from how real users query AI engines:

CodeIntent LayerExample Prompt
DDiscoverability"Best running shoe brands"
PProblem-First"Shoes for plantar fasciitis"
CConsideration"Allbirds vs On Cloud comparison"
EExpert / Authority"What do podiatrists recommend?"
SCustomer Service"Allbirds return policy"
AAsset Intelligence"Allbirds sustainability practices"
LLoyalty / Expansion"New Allbirds releases 2026"
IIdentity / Aspiration"Eco-friendly runner lifestyle brands"
TTemporal / Seasonal"Best summer running shoes 2026"
XEcosystem / Integration"Running shoes compatible with Strava"
RCReputation / Crisis"Allbirds quality complaints"
The 11-layer intent taxonomy — each prompt is auto-classified into exactly one layer based on its phrasing and context.

How Labels Enable Breakdowns

Once every prompt carries labels, all metrics become sliceable:

  • Visibility by intent: Are you strong on Discoverability but weak on Consideration? That tells you buyers find you but don't shortlist you.
  • Visibility by branded vs generic: If branded is 95% but generic is 20%, AI knows who you are but doesn't recommend you to category shoppers.
  • Exclusion by intent: Which buyer journey stages exclude you most? Focus GEO efforts where the gap is largest.

Auto-Derived, User-Editable

Labels are assigned automatically when prompts are created — Branded is derived from your brand name and domain, Intent from the prompt's phrasing and layer code. If the auto-classification is wrong, you can override any label value via the API or dashboard. Reclassifying a label instantly re-segments all historical data — no re-capture needed.