Measures what GPT-5 believes about Perrier from training alone, before any web search. We probe the model 5 times across 5 different angles and score 5 sub-signals.
High overlap with brand prompts shows Perrier is firmly in the model's "sparkling water" category.
Unprompted recall on 15 high-volume discovery prompts, run 5 times each in pure recall mode (no web). Brands that surface here are baked into the model's training, not borrowed from live search.
| Discovery prompt | Volume | Appeared | Positions (5 runs) |
|---|---|---|---|
| What are the best sparkling water brands? | 1,900 | 5/5 | 3, 2, 2, 2, 2 |
| What are the most popular sparkling water options? | 0 | 5/5 | 2, 2, 2, 2, 2 |
| Which sparkling water brands taste the best? | 1,600 | 5/5 | 2, 2, 2, 2, 3 |
| What are the top sparkling water brands to try? | 320 | 5/5 | 1, 3, 2, 1, 2 |
| What are the best sparkling waters for everyday drinking? | 0 | 5/5 | 2, 2, 2, 2, 2 |
| Which sparkling water brands are most recommended? | 0 | 5/5 | 3, 2, 2, 2, 2 |
| What are the best flavored sparkling water brands? | 2,400 | 5/5 | 4, 8, 7, 8, 3 |
| What are the healthiest sparkling water brands? | 140 | 5/5 | 1, 2, 2, 3, 2 |
| What sparkling water brands are worth buying? | 0 | 5/5 | 2, 2, 3, 2, 2 |
| What are the best budget sparkling water brands? | 0 | 0/5 | — |
| What sparkling water brands are most refreshing? | 0 | 5/5 | 2, 2, 2, 2, 3 |
| What are the best sparkling waters for soda replacement? | 0 | 5/5 | 2, 2, 2, 11, 2 |
| What are the best cans of sparkling water? | 70 | 5/5 | 3, 2, 3, 2, 2 |
| What are the best mineral sparkling water brands? | 30 | 5/5 | 2, 2, 2, 3, 1 |
| What sparkling water brands have natural flavors? | 390 | 5/5 | 3, 3, 7, 3, 3 |
This report focuses on Sparkling Waters because that is where Perrier scores highest. The model also evaluates it against the industries below, with their own prompts and competitor sets. Click any industry for its full leaderboard.
Generated automatically from gaps and weaknesses in the analysis above, ranked by potential impact on the AI Visibility Score.
You score 85 on recall but only 37 on retrieval (gap of +47.6). Training-data authority is outpacing your current web footprint. Publish fresh, well-cited content to keep search-augmented responses including your brand.
Close the fragility gapYour LBA is strong. Focus on maintaining authoritative coverage and ensuring new product launches get independent reviews within 12 months of release.
Maintain current LBACore TOM is strong. Watch for specific differentiators (slogans, signature products) that appear in only some iterations. Push those into training-data-crawled headlines.
Maintain / refine TOMOther brands in the Sparkling Waters industry, ranked by overall AI Visibility Score.
Every score on this page is reproducible. Below is exactly what we ran and how we computed each number.
(LBA × Authority × TOM)^(1/3). Geometric mean is used so that any single weak metric pulls the overall score down, rather than being masked by strength elsewhere.
quality × meta × stability × share × recognition × 100. Each sub-signal is on a 0-1 scale. Read the full LBA methodology →
Analysis run on April 23, 2026 at 1:10 AM
Click a prompt to expand its responses. 210 total responses across 72 prompts.