AI Visibility Score: The 3 Dimensions Applied to E-Commerce
Summarize this article with AI
68,000 responses. 100 points. Three dimensions.
68,000 AI-generated responses. That’s Neil Patel’s study ground in fintech. His team turned it into a 100-point AI visibility score.
Three dimensions. Not five, not ten. Three.
Appearance frequency. Early mention. Number of catégories.
The whole thing gives a single score out of 100. Easy to read.
I look at 15 e-commerce sites a week. They all want the same thing: to be cited by ChatGPT, Claude, Gemini, or Perplexity. The question is no longer « Am I visible on Google? » It’s « What place do I hold in the AI response? »
The study covers fintech. The mechanics are the same for a product catalog. An e-commerce brand appears, or it doesn’t. It’s mentioned early, or not. It covers several query catégories, or not.
This audit grid, I reuse it right away. Without waiting for an agency report.
This score measures real presence in the responses buyers already receive, not brand awareness.
A site can rank first on Google. And be absent from an AI response.
That changes the audit.
Fintech is a sector with strong transactional intent. Like e-commerce. The questions are often comparisons or recommendations. « What is the best budget tracker? » becomes « What is the best cordless drill? » The logic stays the same.
Why 68,000? Because responses vary across models, phrasings, contexts. The same query gives a different answer on ChatGPT and on Perplexity. Volume matters.
The grid wasn’t designed for e-commerce, but for fintech. The three pillars still hold anywhere.
In e-commerce, AI assistants become recommendation engines. They recommend a product, a brand, a budget. In these recommendations, it can be measured. It isn’t guessed.
The 100-point score gives a baseline. Frequency, early mention, catégories. I can calculate it myself.
Before talking results, I take a specific example.
Le cas client ne se raconte pas, il se lit. Voici où en était ce site en pièces auto avant tout travail sur sa visibilité IA.
12 points sur 100 : le point de départ
Le score de visibilité IA du client, mesuré sur 47 requêtes, face au palier de 70 jugé décisif.
A 2,300-SKU site, invisible in AI. Here’s what the audit showed.
A client in auto parts. 2,300 SKUs. 1,900 indexed pages. Stable organic traffic at 4,000 sessions per month.
Across 47 AI queries in his market, the starting score: 12 points out of 100.
Seen at this client’s site. Not an estimate.
Appearance frequency was low: 4 responses out of 47. Early mention, poor: no mentions in the top three positions. Catégories covered: 2 out of the 9 expected.
First step, then: map the market’s AI queries. I kept 47, spread across 9 intent catégories. Compatibility, maintenance, budget, troubleshooting, comparison, etc.
Across these 47 queries, the brand appeared in only 4 responses. And always at the end of the response, behind competitors.
The architecture was blocking. Not the content.
I stopped producing isolated pages. I restructured the catalog into semantic cocoons. Each product line got a pillar page, relay pages, targeted internal links.
For early mention, I rewrote page introductions to give the answer immediately. For catégories, I built cocoons around each major compatibility family.
Internal linking connects each product page to its pillar page, each pillar page to its guides, each guide to its comparison pages.
Result: the brand started appearing on queries where it was invisible. First in position 4 or 5. Then in position 2.
Before / After over 14 months
| Indicator | Before | After |
|---|---|---|
| AI score out of 100 | 12 | 68 |
| Organic sessions / month | 4,000 | 37,000 |
| Query catégories covered | 2 | 9 |
| Early mentions (top 3) | 0 | 14 |
Observed result: 37,000 organic sessions per month. +820% over 14 months. The AI score went from 12 to 68.
No ads. No massive content production.
The progression wasn’t linear. It followed the cocoon rollouts. With each covered category, the score gained 5 to 8 points.
This case shows one thing: the 100-point score isn’t a tool quirk. It translates into qualified traffic.
Frequency got corrected by broadening page types: buying guides, comparisons, compatibility pages. Early mention, by rewriting the first 50 words of pillar pages. Catégories, by building them deliberately.
Each lever matches one dimension of the score. That’s what makes the grid actionable.
The three dimensions of the score: frequency, early mention, catégories.
Neil Patel’s study offers a simple grid. The 100-point score combines three indicators.
First indicator: appearance frequency.
Across a panel of queries, how many times does the brand appear in an AI-generated response? That’s the baseline. No presence, no score.
Second indicator: early mention.
Is the brand cited in first position, second, tenth? The earlier the mention, the more it weighs. AI structures its responses to get to the point.
Third indicator: number of catégories covered.
A brand present in 12 query catégories gets more points than one present in 3. That’s the semantic network effect.
The grid applied to e-commerce:
| Dimension | Key question | What it measures |
|---|---|---|
| Frequency | Out of 47 AI queries, how many cite my brand? | Presence coverage |
| Early mention | Does my brand appear in the top 3 positions of the response? | Recommendation strength |
| Catégories | In how many query catégories am I present? | Topical authority |
The fintech study doesn’t detail the exact weighting. In my opinion, that matters little. The most useful thing: measure the three dimensions separately before looking for a single score.
A brand can have correct frequency and zero early mention. It will be cited at the end of the response, behind three competitors. The score reflects it.
In e-commerce, early mention is often neglected. Many audits count Google impressions. Few measure the brand’s position in an AI response.
Frequency alone can mislead. If your brand appears in 30 responses but always in 9th position, the score stays low.
Early mention alone can mislead. If you’re cited first in 3 queries but absent from 44 others, coverage is too thin.
Catégories alone aren’t enough. Being present in 10 catégories with 5% frequency builds nothing.
The score forces you to look at all three axes at the same time.
Early mention changes everything in e-commerce.
In an AI response, the first mention captures attention. The second too. The seventh, much less.
Neil Patel’s study confirms it: early mention weighs heavily in the 100-point score.
Why? AI responses are short. Three to five recommendations. Often fewer.
If your product is cited in fourth position, you exist. But nobody clicks on it.
In e-commerce, the difference comes down to a simple question: « What is the best pair of headphones under €150? »
The brand cited first passes as the main recommendation. The following ones serve as comparison points.
I observed on product pages: moving from 5th to 1st position in AI responses multiplied click-through rate by 2.3.
It’s a reading bias. And it works for you or against you.
On a smart speaker, the voice assistant reads the first recommendation aloud. If it’s not you, the buyer doesn’t hear you.
On a screen, the eye scans the response from top to bottom. The first mention gets the click.
Early mention drives conversion.
How to improve early mention?
- Include the brand name and product in the meta description.
- Place a direct answer to buying questions in the first 50 words of the page.
- Use product structured data with clear brand and model.
- Multiply brand mentions in FAQs and buying guides.
Early mention isn’t obtained by stuffing keywords. It’s built by giving AI an answer that’s easy to cite.
A product page that starts with « Discover our exceptional range » doesn’t help. A page that starts with « Bose QC45 headphones, €149, active noise cancellation » gets cited.
Is your brand cited in the top three positions on the queries that matter?
La troisième dimension repose sur un principe simple : multiplier les contextes. Voici comment les catégories d’un site s’organisent autour de la marque.
L’effet réseau des catégories
Un pillar, vingt clusters. Les trois constellations dorées captent 42 % des citations AI Overview. Survolez pour explorer.
Number of catégories: the semantic network effect.
The third dimension of the AI score is the most underestimated: the number of catégories covered.
A brand present in 12 query catégories gets more points than one present in 3. The study says it clearly.
Why? Because AI builds a representation of the brand. The more it encounters the brand in varied contexts, the more credible it judges it.
In e-commerce, a category equals a purchase intent.
Examples of query catégories for a DIY site:
- Tool comparisons
- Budget buying guides
- Maintenance questions
- Troubleshooting and repair
- Brand compatibility
- New products and innovations
Each covered category strengthens presence in the others. That’s the network effect.
In my deployments, a client moving from 3 to 9 query catégories gains 20 to 45 score points in four months.
An order of magnitude, not a universal promise.
The method: semantic cocoons around each major intent. One pillar page per category, relay pages for sub-intents, internal linking that connects everything.
I apply the DOSE framework taught by Guillaume Attias at BMO Academy. It structures this category expansion without scattering the budget.
Publishing 200 pages is useless. What matters is covering the right catégories with the right architecture.
An e-commerce brand gains nothing from piling up empty catégories. It gains by connecting its existing catégories to precise intents.
AI looks at your product page, yes. It also looks at whether the brand appears in guides, comparisons, technical answers. That cross-cutting presence raises the score.
How many query catégories does your catalog actually cover?
Pas besoin de logiciel payant. La grille de Neil Patel se transforme en méthode en quatre étapes. Suivez le flux.
L’auto-audit en 4 étapes
Un protocole manuel, sans outil payant, pour noter votre visibilité IA.
How to self-audit in 4 steps.
Neil Patel’s grid serves as the basis for a self-audit. Everything is done by hand, no paid tool.
My method fits in 4 steps.
- List 47 AI queries in your market. Mix informational and transactional queries: « best [product] », « how to choose [product] », « [product] reviews », « which [product] for [use] ».
- Query ChatGPT, Claude, Gemini, and Perplexity. Ask each query. Note whether your brand appears in the response.
- Collect three data points per query: presence (yes/no), position of first mention, thematic category.
- Calculate a score out of 100. I weight: 40 points for frequency, 35 for early mention, 25 for number of catégories. That’s my working grid, not Neil Patel’s.
Example log:
| Query | Presence | Position | Category |
|---|---|---|---|
| Best headphones under €150 | Yes | 2 | Comparison |
| How to choose a drill | No | — | Buying guide |
| Which robot vacuum for pet hair | Yes | 5 | Specific use |
The score calculates in three steps. Frequency: number of responses where the brand appears out of 47, scaled to 40. Early mention: proportion of mentions in the top 3 positions, scaled to 35. Catégories: number of catégories covered over the target, scaled to 25.
Let’s take an example. 30 responses present out of 47 give 25.5 frequency points. 12 early mentions out of 30 give 14 early mention points. 6 catégories covered out of 10 targeted give 15 category points. Total: 54.5 points out of 100.
Neil Patel’s official score is something else. This one serves as a compass: it tells you where to focus effort. Frequency, early mention, or catégories.
Below 30 points, AI visibility is almost zero. Between 30 and 60, the brand exists. Above 60, it’s recommended regularly.
The self-audit takes 3 hours. Once per quarter.
Results vary by query and AI engine. Track ChatGPT and Perplexity separately.
The AI score isn’t a gadget. It’s a lever.
The 100-point AI visibility score is a simple évaluation grid. Frequency, early mention, catégories.
Neil Patel’s study shows it across 68,000 fintech responses. The e-commerce transposition is direct.
Frequency tells you if you exist in AI. Early mention, if you’re chosen. Catégories, if you’re a reference.
A brand scoring 20 points out of 100 isn’t dead. It’s poorly structured.
A brand scoring 70 points has a measurable advantage. In traffic, and in trust.
AI doesn’t replace Google. It adds a recommendation layer. Buyers ask questions. Your brand must be in the answer.
The work isn’t to multiply content. It’s to restructure what exists so the three dimensions reinforce each other.
A well-linked catalog, product pages that answer quickly, buying guides that cover the catégories: that’s the foundation.
Neil Patel’s grid gives a unit of measure. It’s up to you to apply it.
I’m not selling you the method. You’ll see the pages.
A serious audit starts with a log. Not with a 40-page report.
You now know what to measure. The question remains whether your brand is cited where it matters.
Across the 47 queries that determine your catalog, in how many does your brand already appear?
Your AI visibility score in 100 points
In the audit, I run your 47 key queries, score frequency, early mention, and catégories, and hand you the grid. You know exactly where to focus your efforts.
Book a strategic call — 45 minFrequently Asked Questions
What is Neil Patel’s AI visibility score?
It’s a score out of 100. It measures a brand’s presence in AI-generated responses. Three indicators come into play: appearance frequency, early mention, and number of catégories covered.
What are the exact three dimensions of the score?
Three things matter: appearance frequency, early mention, and number of catégories. Frequency is the number of responses where the brand appears. Early mention is the position of the first mention. Catégories indicate topical authority.
How do I calculate a 100-point score for my e-commerce?
List 47 AI queries in your market. Query the AI engines. Note presence, position, category. Then weight: 40 points for frequency, 35 for early mention, 25 for catégories. That gives you a compass. The official score is something else.
Is early mention more important than frequency?
No. The three dimensions reinforce each other. High frequency with zero early mention: the score stays low. Strong early mention on few queries: that’s not enough. You have to look at all three axes together.
Where do I start to improve my AI visibility score?
Map 47 AI queries in your market. Log your current score. Then restructure your catalog into semantic cocoons. Result: more catégories covered, and mentions that arrive earlier.
