AI Recommendation Score (ARS)

AI Recommendation Score (ARS) is MencionAI’s proprietary 0–100 index of how strongly AI assistants recommend a brand across monitored prompts and models. Use it for month-over-month tracking: one headline number plus a transparent component breakdown.

Product explainers: What is ARS? · How to read ARS

Why ARS

| Metric | What it answers | | --- | --- | | Plan visibility score | Coverage of mentions/citations for models your plan tracks (Command Center overview). | | Share of voice / mentions | Competitive mix and presence volume. | | ARS | Recommendation strength over time—not only binary presence. |

Formula

ARS is a weighted sum of eight component scores (each 0–100), rounded to an integer in 0–100.

| Component | Weight | API field | | --- | ---: | --- | | Frequency | 20% | frequency | | Position | 15% | position | | Citation quality | 15% | citation_quality | | Model coverage | 15% | model_coverage | | Sentiment | 10% | sentiment | | Share of voice | 10% | share_of_voice | | Prompt breadth | 10% | prompt_breadth | | Consistency | 5% | consistency |

When daily history is thin, consistency is omitted and the other weights are renormalized.

Components (short)

Frequency — Share of analyzed prompt×model cells where the brand is recommended or mentioned.
Position — How high the brand ranks when AI lists recommendations (rank 1 → 100, 2–3 → 70, else lower).
Citation quality — Among presence cells: URL citations and own-domain citations.
Model coverage — Share of planned providers with at least one brand presence.
Sentiment — Normalized tone of brand mentions (neutral baseline when sparse).
Share of voice — Brand mentions ÷ (brand + competitor mentions).
Prompt breadth — Share of distinct monitored prompts with brand presence.
Consistency — Stability of recent daily ARS values.

Where to read ARS

| Surface | Details | | --- | --- | | Product | Dashboard → AI Recommendation Score (/dashboard/command-center/ars) | | Public API v2 | ars_score + ars_components on visibility summary/overview; ars_score on series points | | Public API v1 | Scores are not available (counts only; v1 is frozen) |

Example (v2 summary fields):

{
  "ars_score": 62,
  "ars_components": {
    "frequency": 70,
    "position": 55,
    "citation_quality": 40,
    "model_coverage": 80,
    "sentiment": 58,
    "share_of_voice": 45,
    "prompt_breadth": 60,
    "consistency": 90
  }
}

Interpreting ARS

Prefer multi-week / MoM trends over single-day spikes.
When the headline moves, inspect components—frequency vs position vs SoV tell different GEO stories.
Keep the prompt set stable when comparing periods.
Treat ARS and plan visibility score as complementary (strength vs coverage).

AI Recommendation Score