04 · Understand
Recommendation Intelligence™
Does AI actually recommend your brand?
A category-level report showing exactly who AI recommends, how often, where they appear, and what separates them from your brand across buyer intents.
$999 one-time report
How this is measured. We ask an AI model 20 high-intent buyer questions in your category, each run 20 times to capture variance, for 400 recommendation observations. For every answer we record which brands were recommended, in what position, and the reason the model gave.
Example result · illustrative, activewear category
When shoppers ask AI what activewear to buy, this brand is named in 15% of conversations. It is completely absent from the other 85%.
The brands AI recommends ahead of it, Nike, Under Armour, and Adidas, do not win on product quality. They win because AI understands them as performance brands. AI understands this brand as a fashion brand, so it quietly routes every performance question elsewhere.
This is not a visibility problem that more ads or SEO will fix. It is a recommendation problem, and the fix starts with language, not the website.
14.8%
Recommendation Share
5 / 20
Buyer questions present
4.7
Avg. position when shown
The competitor landscape
Who AI recommends in your category
Recommendation Share across all 20 buyer questions, with the average position each brand holds when it appears.
Example result · illustrative, activewear category
| # | Brand | Recommendation Share | % | Avg pos |
|---|
| 1 | Nike | | 57.3 | 2.4 |
| 2 | Under Armour | | 51.3 | 4.5 |
| 3 | Adidas | | 47.3 | 3.7 |
| 4 | Reebok | | 42.5 | 7.3 |
| 5 | Lululemon | | 35.8 | 2.3 |
| ··· |
| 15 | This brandYou | | 14.8 | 4.7 |
Recommendation Positioning Map
The category, mapped
Higher means AI recommends the brand more often. Further right means it shows up across more buyer questions. The strong brands cluster top right.
Example result · illustrative, activewear category
Recommended more often ↑
Appears in more buyer questions →
Nike
Adidas
U.Armour
Reebok
Lulu
Gymshk
Athleta
N.Bal
You
Competitors AI recommends
This brand
Bubble size = how many buyer questions the brand appears in
Competitive language gap · most actionable
AI describes the winners and your brand in completely different words
These are the actual reasons the AI model gave when recommending each brand. This is the model's own language, not our interpretation.
Example result · illustrative, activewear category
What AI says about the winners
moisture-wicking technology
cushioning & support
durability
stability for heavy lifts
performance fabrics
built for intense workouts
What AI says about this brand
stylish
fashionable
chic
comfortable for casual wear
perfect for yoga
fashion-forward
AI has filed this brand under fashion, not performance. That one framing decides everything that follows. It wins the lifestyle questions and disappears from every performance question, because AI does not connect it to that language at all.
The blind spots
Buyer questions where your brand never appears
High-intent questions where the brand was recommended zero times across 20 runs, and the brand AI named first instead.
Example result · illustrative, activewear category
best gym wear for menAI's #1: Nike
best moisture wicking shirtsAI's #1: Under Armour
best fitness trackerAI's #1: Apple
best adjustable dumbbellsAI's #1: Bowflex
best home gym equipmentAI's #1: Peloton
What predicts a recommendation, and what does not
Most AI visibility tools are measuring the wrong thing.
Most tools score site technical quality: schema markup, page speed, structured data. We tested whether any of that predicts if AI actually recommends a brand.
0.7%
correlation between site technical quality and being recommended
68.9%
of recommendations change just from swapping the underlying AI model
46.1%
of brands still differ asking the identical question twice, the noise floor every claim has to clear
61.4%
rank stability month over month, once a brand is a repeat winner
Fixing a website will not fix this on its own. Which AI model answers, and whether it is browsing the live web, moves the outcome far more than a technical audit can touch. Every number in this report is built from repeated runs, not one lucky or unlucky query.
Cross-category context
This pattern is not specific to one category.
Across 10 categories and roughly 40,000 measured recommendations, a small set of famous names dominates, and everyone else fights over the remainder. Below is how much of each category the single top brand captures.
Example result · illustrative, cross-category benchmark
Fitness & Activewear
57.3%
The winners are always the famous names: Apple in electronics, Nike in fitness, Everlane in fashion. The more concentrated the category, the harder it is for any brand to enter it.
Where to start
Three moves the data actually supports.
Not a guarantee that a number moves by a fixed amount. The three places this kind of dataset points to first, in priority order.
01
Close the language gap
Rewrite the product and category pages your brand controls to use the vocabulary AI already rewards in your category, instead of style-led language alone.
02
Answer the missing buyer questions directly
Publish content and product pages that speak plainly to the buyer questions your brand is currently invisible on. AI tends to recommend what it can clearly match to the question being asked.
03
Track across models and conditions
Results shift by up to 68.9% when the model changes. Track recommendations across models and search conditions on a schedule, not one afternoon's chat.