Presence rate
Answers mentioning the brand / all eligible answers.
Requires: Entity coding
What Agent Wants
All WAW metrics
Understand how AI represents your brand: visibility, description and accuracy, sources, and competition. ASoV is one weighted-share metric within this wider analysis.
A metric appears only when the required evidence exists. Missing coding, citations or verified facts produce an unavailable value, never a zero. Some detailed views are still being prepared.
Visibility and recommendation
Presence, first mention, recommendation, weighted share, position and prompt coverage.
First mention
Answers naming the brand first / all eligible answers.
Requires: Entity coding
Recommendation rate
Answers explicitly recommending the brand as primary / all eligible answers.
Requires: Role coding
Agent Share of Voice
Weighted brand contributions / contributions from all brands. Role and textual rank determine the weight.
Requires: Role coding
Average position
Sum of distinct-brand positions / answers mentioning the brand.
Requires: Entity coding
Prompt coverage
Model/prompt/design combinations with a presence / all eligible combinations.
Requires: Entity coding
Primary recommendation among presences
Primary recommendations / answers mentioning the brand. This is not purchase conversion.
Requires: Local map prototype
Description and accuracy
Attributes, sentiment, verified claims, errors and entity confusion.
Sentiment
Mean negative (−1), neutral (0) and positive (+1) observations; mixed sentiment is shown separately.
Requires: Sentiment coding
Observed attributes
Answers associating an attribute with the brand / answers mentioning it.
Requires: Attribute coding
Desired and observed attributes
Declared desired attributes alongside their measured association rates; no invented alignment score.
Requires: Desired attributes and attribute coding
Accuracy
Correct claims / verifiable claims. Unverifiable claims are excluded and reported separately.
Requires: Dated GroundTruth
Incorrect claims
Incorrect claims / verifiable claims.
Requires: Dated GroundTruth
Obsolete claims
Obsolete claims / verifiable claims.
Requires: Dated GroundTruth
Error severity and category
Verified errors by category and severity / verifiable claims.
Requires: Dated GroundTruth
Entity confusion
Answers with verified entity confusion / answers containing verifiable claims about the brand.
Requires: Dated GroundTruth
Sources and citations
Native citations, official-domain share and source mix.
Citation share
Native citations to verified official domains / all native citations in documented web-enabled answers.
Requires: Captured native citations
Source mix and domains
Citation counts by source type and domain, with distinct supporting answer counts.
Requires: Captured native citations
Competition and actions
Explicit comparisons, co-presence, substitution and stability.
Head-to-head win rate
Explicit wins against a competitor / answers mentioning both brands; ties and unknowns remain separate.
Requires: Explicit comparison coding
Average head-to-head win rate
Equal-weight mean of evaluable competitor win rates; incomplete pairs prevent a complete summary.
Requires: Explicit comparison coding
Co-presence
Answers mentioning both brands / answers mentioning the focus brand.
Requires: Entity coding
Substitution rate
Branded-focus questions recommending another brand as primary / all eligible branded-focus questions.
Requires: Branded questions and role coding
Recommendation stability
Matching primary-brand sets / pairs of true repetitions of the same model and prompt.
Requires: Repeated prompts
Mean and dispersion
Equal-weight mean and population standard deviation of selected models’ ASoV, using a complete paired design.
Requires: Local map prototype
Agent Gap Matrix
log₂(AI share / market share), only for positive compatible shares. No sales effect is implied.
Requires: Compatible external market dataset
Sample and uncertainty
Every result needs its denominator, sample and limitations.
Data quality and uncertainty
Eligible, coded, unresolved and failed responses; 95% intervals, sample sizes and explicit reasons for missing values.
Requires: Always inspect alongside each metric
Commerce, actions and tracking
Operational views build on the same analysis and preserve its evidence.
Commerce signals
Product qualities, retailers, written URLs and cited sources in saved answers. Counts describe those answers, not transactions.
Requires: Available extraction; legacy method is identified
Action plan and tracking
Evidence-linked tasks and comparable repeated benchmarks. Budgets and expected impacts are planning assumptions.
Requires: No causal impact or sales measurement