About AgentRank
Search has compressed into a single answer. Shoppers no longer scroll a results page — an LLM hands them a recommendation. The brand named in that answer wins discovery, consideration and the click. The ones omitted lose share quietly, model by model, market by market.
AgentRank is the measurement layer for that new surface. It is a continuous audit of how leading AI models talk about your category: where your brand stands, which retailers and content ecosystems each model leans on, and what to do next.
What it means for you
If you aren't managing how AI platforms recommend your brand, someone else will set the narrative. AgentRank tells you which models drop you, in which segment, and what to do next.
Three Intelligence Modules
AgentRank is organised into three modules. Each answers a different question. Together they give growth teams a complete picture of AI Share of Voice — from brand-level exposure to the information ecosystems that drive each model's answers.
Section 01
Brands
Defend share, brand by brand.
- —AI Share of Voice across every tracked brand
- —Model consensus leader — who AI platforms agree on as #1
- —Active erosion alerts — brands slipping out of top-5
- —Opportunity radar — auto-generated next moves
Section 02
Manufacturers
Roll up the house of brands.
- —Parent-company AISOV rollup across all owned brands
- —Portfolio drill-down — segment strength, market split
- —Cross-platform coverage side-by-side
- —Best-rank watch by model at a glance
Section 03
Platforms
Where the answer comes from.
- —Most influential retailers cited per market & category
- —Information ecosystems — social, video, forums ranked
- —Per-platform phrase bank & deep dive
- —Non-negotiables — what to fix when all models agree
What Is the Dataset
AI assistants are becoming a discovery channel. Brands that appear first in AI responses gain preferential exposure at the moment of intent — with no ability to buy their way in through advertising.
Atomic unit
Each row in the dataset = one brand recommendation made by one AI platform, in one country, for one product segment, at one point in time.
How to Read a Row
A single row carries five layers of context — when it was captured, what was asked, where, by which AI, and what came back.
| Layer | Example |
|---|---|
| Snapshot | SNAP-2026-03-M01 · March 2026 · Final |
| Segment | Beauty & Personal Care → Skin Care → Facial Care: Cleansing |
| Geography | United States · AMER · Developed · English |
| AI platform | ChatGPT (GPT-5) · General AI |
| Recommendation | CeraVe · Rank 1 · Score 35 · L'Oréal · Mainstream · Leader |
Field Reference
The dataset contains 37 fields across five logical groups.
A
Group A — Snapshot & Run Identifiers
Identify when and how the data was collected. Use these to filter by wave and track changes over time.
Snapshot_ID
Unique identifier for the collection run (e.g. SNAP-2026-03-M01).
Snapshot_Date
Date the data was captured (Excel serial date).
Snapshot_Period
Start of the observation window.
Frequency
How often this data is collected — Quarterly.
Category_Batch
Human-readable run label (e.g. 'March Skincare WIP Run').
Data_Status
WIP or Final. Never share WIP externally.
Notes
Free-text annotation describing scope of the snapshot.
B
Group B — Category & Segment
Define the product universe. A segment is the specific sub-category posed to the AI — the more granular, the better the competitive read.
World
Broadest umbrella — Beauty & Personal Care.
Category
Primary product category (Skin Care, Hair Care, etc.).
Segment_Code
Stable alphanumeric code (BPC-SC-1) — use for joins.
Segment_Name
Human-readable segment (e.g. 'Facial Care: Cleansing').
Prompt_Type
Persona used when querying — 'General Shopper' is baseline.
C
Group C — Geography & Market Context
Where the query was run and the market characteristics of that geography. Explains why rankings differ between, say, France and Germany in the same segment.
Country
Specific country the AI was prompted for.
Region Run
Macro region (e.g. NA & Europe).
Greater Region
AMER, EMEA — high-level grouping.
Region
North America, Europe — finer grouping.
Market Tier
Developed or Emerging.
Language
Localised query language.
Market Maturity
Mature or Growth.
Digital Commerce Strength
High / Medium / Low.
Social Commerce Strength
High / Medium / Low.
Premiumization Level
Tendency to trade up to premium.
D
Group D — AI Platform
Which AI system made the recommendation. Rankings differ significantly between platforms — this is the core comparative dimension.
Platform
ChatGPT, Gemini, Claude, Grok, Deepseek.
Model Version
Specific model version queried (e.g. GPT-5, Gemini 2.5).
Platform Type
General AI or Social AI (e.g. Grok on X).
E
Group E — Brand & Recommendation Output
The core output fields — what the AI actually recommended and how the brand is characterised.
Brand_Name
Recommended brand (e.g. CeraVe, La Roche-Posay).
Brand ID
Stable internal code (BPC-1) — always join on this.
Manufacturer_Name
Parent company (L'Oréal, Unilever, Beiersdorf...).
Manufacturer ID
Stable manufacturer code (MF-000001).
Rank
Position 1–5 in the AI's response. Lower = more visible.
Score
Position-weighted points: 35 / 25 / 20 / 12 / 8.
Price Tier
Value, Mainstream, Premium.
Positioning
Leader, Challenger, Specialist, Niche.
Market Footprint
Global, Regional, Local Champion.
Revenue Tier
Mega (>$20B), Major, Large, Mid-Sized, Independent.
HQ
Headquarters country.
Revenue M USD
Approx. annual manufacturer revenue (USD millions).
Scoring & Methodology
Scores are position-weighted, not raw counts — they reflect the relative visibility premium of each rank position.
Rank 1
35
Highest — first mention
Rank 2
25
Strong — second
Rank 3
20
Visible — third
Rank 4
12
Marginal — fourth
Rank 5
8
Minimum — fifth
AI Share of Voice formula
Sum a brand's Scores across all platforms for a given segment and country, divide by the maximum possible (5 platforms × 35 = 175).
Example: CeraVe scores 35 (ChatGPT R1) + 25 (Claude R2) + 20 (Gemini R3) = 80 / 175 = 45.7% AISOV in US Facial Cleansing.
Example: CeraVe scores 35 (ChatGPT R1) + 25 (Claude R2) + 20 (Gemini R3) = 80 / 175 = 45.7% AISOV in US Facial Cleansing.
Common Filters & Use Cases
| Use case | Filter / method |
|---|---|
| Focus on one brand across all markets | Filter Brand ID = [ID] (more stable than name) |
| Compare platforms in one country | Filter Country, pivot Platform × Rank |
| Identify Rank 1 sweeps | Filter Rank = 1, count unique Brand by Segment |
| Track one segment across countries | Filter Segment_Code, group by Country + Platform |
| Isolate premium brands only | Filter Price Tier = Premium |
| Exclude WIP data | Filter Data_Status = Final |
| European markets only | Filter Greater Region = EMEA |
| Score share by manufacturer | Group by Manufacturer, sum Score / total segment pool |
Glossary
AI Rank
Position (1–5) where a brand appears in an AI assistant's response.
AI Share of Voice
Combined score across platforms in a segment/country, as % of maximum.
Snapshot
A single collection run, identified by Snapshot_ID.
Segment
A specific product sub-category used to frame the query.
Prompt_Type
Persona used when querying. 'General Shopper' is baseline.
Score
Position-weighted points: 35 / 25 / 20 / 12 / 8.
WIP
Work In Progress — not yet finalised. Do not share externally.
Market Footprint
Global, Regional, or Local Champion.
Positioning
Leader, Challenger, Specialist, or Niche.
Price Tier
Value, Mainstream, or Premium.
Platform Type
General AI vs. Social AI (e.g. Grok on X).
Brand ID / Manufacturer ID
Stable internal codes — preferred over names for joins.
Questions about this dataset? Contact your ConsensysAI account team.