Case Studies
Real examples of how brands improved their AI visibility and started appearing in ChatGPT, Gemini, and Copilot recommendations.
The Best Product Data in Baby Sleepwear. Zero Structured Data Means AI Agents Cannot Find It.
Kyte Baby has the richest descriptions in the baby category and a devoted community. Zero JSON-LD means the data is trapped. 4% AI visibility.
922 Reviews on One Product. AI Has Never Mentioned the Brand.
Lalo's Suction Bowl has 922 reviews at 4.6 stars. Their Play Kitchen costs $740. Product descriptions are 12-32 words. Zero AI visibility across 150 tests.
aden + anais Is the CeraVe of Muslin Blankets. Little Unicorn Is Completely Invisible.
Every muslin query on every AI platform defaults to aden + anais. Little Unicorn makes beautiful muslin blankets with 30-word descriptions and 1 tag per product. Zero AI visibility.
One Product Advantage, One Platform, One Query. That Is the Entire AI Footprint.
Mushie surfaces 90% of the time on Gemini for mold-free bath toys. On every other query, every other platform — zero. The most concentrated visibility profile in the audit set.
The Most Loved Baby Eczema Product in Parenting Communities. Invisible to Every AI Agent.
Tubby Todd's All Over Ointment is a cult product. Parents swear by it daily in Facebook groups and Reddit. 1% AI visibility. Pharmaceutical brands own every eczema query.
Solid foundation. Internal-only tags. The niche specialism that doesn't exist in structured data.
Act+Acre is built on scalp care specialism — exactly the kind of niche AI agents should reward. The basic data infrastructure works. But the tags don't reflect the niche, so the brand competes for scalp queries with no structural advantage over generalists.
2 million YouTube subscribers. 0% AI visibility.
Beardbrand essentially created the modern premium beard care movement through a decade of YouTube content. AI agents recommend Amazon-native Honest Amish and Viking Revolution instead. YouTube subscribers are not an AI signal.
41 tags. 404 errors. The best taxonomy in the audit set lives on broken pages.
Ceremonia's Aceite de Moska carries 41 tags with a structured prefix taxonomy — Ingredient_, Benefit_, product-badge_. The best tag system across 55 audits. Two of three audited products return 404 errors. World-class data sitting on broken catalogue infrastructure.
30 years of ethical sourcing. 14% AI visibility.
Counter Culture owns 'ethically sourced coffee' on Copilot at 80% — but its 30-year reputation can't compensate for thin product data. Identity is a real AI signal, but it's a ceiling, not a foundation.
Zero words. $38 price tag. The Vogue-level brand with a white-label data profile.
Crown Affair has Vogue features, a cult following, and products priced at $38-42. One product literally returns an empty description field. The structured data profile is the thinnest recorded across 55 brand audits.
Buy-one-give-one mission. 5% AI visibility.
DIFF has the most extensive tag taxonomy in the eyewear group and a charitable mission that differentiates it from hundreds of fashion sunglasses brands. The mission appears nowhere in product data — and AI agents recommend Arlo Wolf instead.
23 tags. Half promotional. Colour-specific data drowned out by discount noise.
dpHUE's Cool Blonde Conditioner carries 23 tags — the highest in the hair care audit group. Half are attribute tags that genuinely help AI agents. The other half are bogo-free and promo_25%-off. AI agents process them all as product metadata.
12 tags per product. Zero describe the product.
Dr. Squatch carries 12 tags per product — a healthy count. Every single tag is internal: edition_core, exclude_rebuy, subscription-eligible_yes. Not one describes scent, skin type, or ingredients. The result: 0% ChatGPT visibility.
1.7 stars on Trustpilot. 100% on Copilot. Fellow's AI paradox.
Fellow's Stagg EKG is the category default for pour over kettles on Copilot — 10/10 runs at position one. The Opus grinder is invisible. Same brand, same data infrastructure, completely different outcomes.
Floyd sells a $2,450 bed. Its product data says it is made by RIZE.
Every Floyd product carries the vendor name 'RIZE' — apparently a fulfilment partner — in the data field AI agents read as brand identity. Combined with missing schema and a 1.6/5 Trustpilot, Floyd is one of the most structurally disadvantaged brands in the home decor audit set.
27,769 reviews. No product schema. The most validated brand with the weakest data infrastructure.
Function of Beauty has 27,769 Trustpilot reviews at 4.0 — roughly 10x more than any other brand in the audit set. Two of three audited products return no product schema. The most validated brand is structurally invisible to AI agents on most of its product pages.
Recycled Water Bottles, XXS to 6XL, and a Data Layer That Tells AI Agents None of It
Girlfriend Collective has the widest size range (XXS to 6XL) and leggings made from recycled water bottles. Neither differentiator appears in the product data AI agents consume.
Zero schema. Zero reviews. A house brand that doesn't exist in its own product data.
Hawkins New York is known in the design world for its curation and minimalist aesthetic. Two of three audited products carry the vendor name 'Sophie Lou Jacobsen' instead of Hawkins New York. The house brand identity that exists for human shoppers has no data representation.
869 reviews. 18-word descriptions. The most fixable AI gap in grooming.
Huron has the best review infrastructure in the grooming audit group — aggregateRating on every product, hundreds of reviews each. And the thinnest descriptions. AI agents can verify the products are well-rated but have almost nothing to recommend them for.
What 286 Words of Product Content Teaches About Winning AI Recommendations
In a category where 10-40 word descriptions are the norm, Janji's 286-word product pages are a structural competitive advantage. A case study in what good looks like.
35 tags per product. 0% AI visibility.
KREWE's CONTI Haze 12K has 35 tags — the most comprehensively tagged product across 100 brand audits. Zero appearances on any platform for any query. Tags without editorial authority are metadata that nobody reads.
A $908 coffee table. 36 words of description.
Lulu and Georgia is built on designer curation — interior designers collaborate on the selection, customers buy into the taste. AI agents do not experience curation. They experience data. And the data on a $908 coffee table is 36 words.
66-word flagship. 9-word corrector. Subscription brands need consistency most.
Lumin's product descriptions range from 9 to 66 words — the widest content variance in the grooming audit group. For a subscription brand whose value proposition is the range, fragmented data means AI agents see disconnected products, not a system.
Best product description in the coffee group. 1% AI visibility.
Onyx Coffee Lab has SCA competition wins, the richest product descriptions in the coffee audit set, and zero JSON-LD on any product page. The result: 1% AI visibility — the most extreme reputation-to-visibility disconnect in the entire 55-brand audit.
The Always Pan went viral on TikTok. Its description is 5 words.
Our Place became one of the defining DTC launches of the 2020s through TikTok virality and Selena Gomez. The Always Pan — a $199 product — has a 5-word product description. TikTok doesn't feed AI agents.
The Brand That Created a Movement and Left Nothing for AI Agents to Read
Outdoor Voices raised $60M and defined a generation of activewear. Zero product schema. 19-word descriptions. 2.3-star Trustpilot. When culture moves on, what remains is what you put in the data.
Zero words on the product page. 52% Copilot visibility.
Pair Eyewear's product pages contain zero words of body description. Copilot surfaces the brand 52% of the time anyway. Category creation is the most powerful AI visibility signal we've measured — but it has limits.
Zero descriptions. Zero schema. 1.7 stars. The data floor.
Across 55 DTC audits, Parachute has the weakest product data of any brand audited. Three products tested, zero words of description on all three, zero JSON-LD on any product, 1.7/5 Trustpilot with 411 reviews. The data floor.
The Rapha of Running Has No Data Architecture
Satisfy is to running what Rapha is to cycling. Paris-based. Fashion-credible. Performance-legitimate. Zero product schema. 39 internal tags. No AI agent can read why a $210 running short is worth $210.
141 words. Zero structured data. 3% AI visibility.
Sunski writes the richest product descriptions in the eyewear audit group — three times longer than DIFF or Tens. Without JSON-LD wrapping that content, AI agents have no structured path to it. The best product stories in the group are wrapped in the worst infrastructure.
Award-winning niche razor. _label_Award-Winning is not a product attribute.
Supply makes premium single-edge razors with award-winning design. The award appears as the internal tag _label_Award-Winning. No AI agent will parse that as a quality differentiator. Niche brands cannot win on volume — they have to win on specificity.
The Compounding Problem: When Bad Reviews Meet Missing Data
606 Trustpilot reviews at 1.5 stars. Zero product schema. 10-word descriptions. When every data layer is weak simultaneously, there is nothing to compensate for anything else.
Owns 'warm filter sunglasses.' AI agents recommend Oakley Prizm instead.
Tens makes lenses that apply a warm, film-like tint to everything you see — a genuinely unique product. Three products share an identical 44-word description. AI agents cannot tell the warm-tint lifestyle product apart from Oakley Prizm sport performance lenses.
Best Trustpilot in the coffee group. Worst AI visibility.
Trade Coffee is 4.6/5 on Trustpilot — the best in the coffee group. AI visibility: 3%. The taste quiz that powers the matching algorithm collects exactly the data AI agents need, but the product pages they crawl are nearly empty.
15 tags per product. 0% on product queries.
Verve has the highest tag count of any brand in the coffee audit group and owns 'California coffee roaster' on Copilot at 90%. Every product-level query returns zero. More tags doesn't mean more visibility — quality and query mapping do.
When AI Agents Know Your Differentiator But Still Won't Recommend You
Ana Luisa has been Climate Neutral Certified since 2020. AI agents know it. They recommended it 4% of the time anyway.
Structured Data Is Not Enough. A Pet Brand Audit That Proves It.
Fable has the best JSON-LD of any pet brand tested. 9/10 structured data score. 7% AI visibility — the lowest in the pet audit set.
The Best Structured Data in Jewellery, Filled with 32 Identical Words
Gorjana has excellent JSON-LD, 4.9-star ratings, and thousands of reviews. Every product uses the same 32-word template description. 9% AI visibility.
The Queen of the Hoop That AI Agents Won't Recommend
Gemini can recite Jenny Bird's tagline. Celebrity fans include Kylie Jenner and Hailey Bieber. Ask any AI for the best chunky hoop earring, and Jenny Bird surfaces 1 out of 30 times.
220 Words of Product Copy. Zero Structured Data. 2% AI Visibility.
Maxbone has the best product descriptions in the pet category. And the worst AI visibility. 220 words of rich copy trapped behind zero structured data.
Gemini Says 'The Gold Standard.' Copilot Has Barely Heard of It.
Mejuri has 86% AI visibility on Gemini and 24% on Copilot. Same brand, same products, same day — three completely different realities.
62% on Gemini, 0% on ChatGPT — The Brand Fashion Editors Love That AI Cannot Find
Missoma is 'the current it-brand for layering' on Gemini and does not exist on ChatGPT. Celebrity endorsements and editorial credibility — with product data that tells AI agents nothing.
0% Across All Platforms — Total AI Invisibility for a Premium Pet Brand
WagWear makes premium dog leashes with Sunbrella cording and brass hardware. 150 automated tests. Zero appearances. Total AI invisibility.
Copilot Recommends This Brand 100% of the Time. ChatGPT Has Never Heard of It.
West Paw has 74% visibility on Gemini, 100% on Copilot's eco-friendly query, and 0% on ChatGPT. The widest platform gap measured.
The Top Design Dog Brand on Gemini. It Does Not Exist on ChatGPT.
Wild One surfaces at position #1 on Gemini for stylish dog leashes. It does not exist on ChatGPT. The brand's design reputation outpaces its product data.
A Shark Tank brand with a 34-word product description. And it is the #2 most visible brand in the audit set.
Bala achieves 40% AI visibility — second-highest in the audit set — with Gemini at 78% and Copilot ranking it #1 for ankle weights. The product description is 34 words.
Brightland is a design icon that two AI platforms have never heard of
Brightland achieves 14% AI visibility across 150 tests — but 100% of it comes from Gemini. ChatGPT and Copilot have never heard of this California olive oil brand.
The highest-scoring DTC brand we have tested. And ChatGPT still returns zero.
Caraway achieves 41% AI visibility — highest in the audit set — with an extraordinary 82% on Gemini. But ChatGPT returns 0% across 50 tests. Same brand, same day, three different realities.
12 million YouTube subscribers. 3% AI visibility.
Chamberlain Coffee achieves just 3% AI visibility across 150 tests — the lowest score in the entire audit set. Celebrity fame does not equal AI product visibility.
The brand that broke ChatGPT's 0% barrier
Fly By Jing achieves 27% AI visibility — the highest in the audit set — and is one of only two brands to break through ChatGPT. Cultural penetration beats product data.
The brand that created squeeze bottle olive oil. And ChatGPT has never heard of it.
Graza achieves 17% AI visibility — strongest on its defining format query where Gemini calls it 'The Gold Standard'. But ChatGPT recommends OXO dispensers instead.
Known but not recommended. The AI visibility gap between awareness and authority.
Material Kitchen achieves 28% AI visibility — above average — and is one of only two brands to break ChatGPT's 0% barrier. But it consistently ranks 4th or 5th, never first.
One query. That is this brand's entire AI presence.
Muddy Bites achieves 10% AI visibility — but all 15 surfacings came from a single query. If you don't already know waffle cone bites exist, AI will never introduce you.
$200M revenue. 30,000 retail locations. Zero ChatGPT visibility.
Olipop achieves 31% AI visibility — highest in Food & Beverage — despite having the worst product data in the audit set. Category authority compensates for data gaps, but ChatGPT's 0% shows the limits.
The brand that AI knows by name but never recommends
Open Spaces achieves 13% AI visibility — but 100% on Gemini for DTC brand queries and near-zero for actual product queries. The DTC storefront appears shut down.
Poetry vs. Product Data: Why Boy Smells Is Invisible to AI Shopping Agents
We audited Boy Smells, a gender-neutral candle brand sold at Nordstrom and Saks Fifth Avenue with coverage in Vogue and GQ. Across 150 tests, they surfaced just 12% of the time. ChatGPT recognises them as an indie brand (60% surfacing). Gemini barely knows they exist (4%). Their core differentiator — a coconut and beeswax blend — appears nowhere in their product descriptions.
I Audited Brooklyn Candle Studio Against AI Shopping Agents. Dominant on Two Platforms, Invisible on the Third.
We audited Brooklyn Candle Studio, a clean candle brand operating from Industry City, Brooklyn. Across 150 tests, they surfaced 15% of the time. But 74% of those surfacings came from a single query — Brooklyn eco-friendly candle. On that query, Gemini and Copilot put them at #1 (90% and 80%). ChatGPT? 0/50. Zero. The most polarised visibility profile in the audit programme.
Zero structured data. Gemini calls the Allyn Tote 'the gold standard.' ChatGPT and Copilot have never heard of them.
Dagne Dover has zero JSON-LD — a first in 30 audits. Gemini carries 100% of their AI visibility at 52%, calling the Allyn Tote 'the gold standard' for neoprene. ChatGPT and Copilot: 0/100 combined. The most extreme platform disparity in the audit programme.
Perfect Product Pages, Zero AI Visibility: The Earl of East Paradox
We audited Earl of East, a London-based candle brand with the best structured data in the audit programme — aggregateRating on every product, 264-word descriptions with room size guides. Across 150 tests, they surfaced 11% of the time. ChatGPT: 0%. Copilot: 0%. Gemini: 32%. The most extreme platform disparity in 22 brand audits.
Zero words. Every product description is empty. 25% AI visibility.
Horizn Studios has literally zero words on every product page — yet achieves 25% AI visibility driven entirely by editorial coverage on Gemini. The paradox of brand without data.
The best product tags in 30 audits. The worst structured data. ChatGPT: 0/50.
Keychron has the highest AI visibility in the audit set at 45% — Gemini 80%, Copilot 56%. But ChatGPT has never heard of them. The best product taxonomy (19 structured Key:Value tags) paired with the worst structured data (no Product schema). A $249.99 keyboard described in 52 words of packing list.
A $1,099 suitcase described in 94 words
LEVEL8 has genuine specs — German Makrolon polycarbonate, YKK zippers, RPET lining — buried under messy formatting, useless tags, and zero editorial presence. 17% AI visibility.
The best product descriptions we have ever audited. ChatGPT recommended them zero times out of 50.
Monos has 300-460 word descriptions with exact specs, 4.9/5 ratings, and comprehensive structured data. ChatGPT returns 0/50. The gap between data quality and AI visibility.
The best product data in tech accessories. 13% AI visibility.
Native Union scores 9/10 on descriptions and 9/10 on structured data — the highest in tech accessories. But actual AI visibility is just 13%. 0% on ChatGPT. Rich product pages are necessary but not sufficient.
0% on ChatGPT. 74% on Gemini. Same brand. Same products. Same day.
Nomad is the #1 recommendation for leather iPhone cases on Gemini — 10 out of 10 runs. ChatGPT has never mentioned them. 0 out of 50 tests. The widest gap between brand reputation and product data in the audit set: Wirecutter, The Verge, MKBHD coverage — and a 24-word MacBook sleeve description.
Gemini calls it 'the gold standard.' The product_type field is empty on every product.
Ohsnap is NOT invisible to AI — Gemini recommends it 46% of the time at #1, calling it the 'gold standard' for thin phone grips. But ChatGPT returns 0/50. The product_type field is empty on every product. The most basic categorisation field in Shopify — blank.
I Audited P.F. Candle Co. Against AI Shopping Agents. 23 Words. 909 Reviews. The Highest AI Visibility in the Audit Set.
We audited P.F. Candle Co., a brand sold at Target and Nordstrom with 909 reviews on a single product. Across 150 tests, they had the highest visibility in the candle audit set at 22%. But ChatGPT returned 0/50. Their product descriptions are 23 words. The word 'soy' does not appear once. Reviews and editorial presence outweigh description quality — but only on some platforms.
I estimated 33%. The real number? Also 33%. But I got everything else wrong.
Satechi's overall 33% matched the estimate exactly — but the per-platform reality was completely inverted. ChatGPT was estimated at 50% and scored 0/50. Copilot was estimated at 30% and scored 38%. The aggregate was right. The story was wrong.
55,000 reviews. 26 words of product description.
Stubble & Co claims 55,000+ reviews but gives AI agents 26 words to work with. Zero tags on hero products. No dimensions, no weight, no airline compliance. Near-invisible.
I Audited Apotheke Against AI Shopping Agents. The Best Description Template in the Set, Zero Social Proof.
We audited Apotheke, a Brooklyn-based premium candle brand stocked at Bloomingdale's and Nordstrom. They have the best product description template of any brand in this audit programme — structured sections, scent pyramids, burn times, dimensions. Across 150 tests (10 repeats per query per platform), they surfaced just 14% of the time. The template is perfect. Everything around it is broken.
I Audited a Harley Street Dermatologist's Brand Against AI Shopping Agents. 8,000 Reviews. Zero Recommendations.
We audited Dr Sam's, a Harley Street dermatologist's skincare brand with 8,000 five-star reviews and JSON-LD structured data. They surfaced on 0 out of 15 AI shopping queries. The widest gap between data infrastructure and actual visibility we've seen.
I Audited Herbivore Botanicals Against AI Shopping Agents. A $50M Clean Beauty Brand. Zero Recommendations.
We audited Herbivore Botanicals, a $50M clean beauty brand stocked at Sephora and Ulta. They surfaced on 0 out of 15 AI shopping queries. But the real finding is bigger: clean beauty has no AI consensus. Every platform recommended completely different brands. The category is wide open.
I Audited Summer Fridays Against AI Shopping Agents. Sephora's Best-Seller Is Almost Invisible.
We audited Summer Fridays, the influencer-founded brand behind Sephora's best-selling Jet Lag Mask. 1.1 million Instagram followers. Best structured data in our audit set. They surfaced on 2 out of 15 AI shopping queries. One of those was from eBay.
I Audited UpCircle Beauty Against AI Shopping Agents. The First Brand That Actually Works.
We audited UpCircle Beauty, a B Corp-certified upcycled skincare brand. They surfaced on 8 out of 15 AI shopping queries — the highest score in this audit programme. When the query matched their sustainability story, AI agents put them first. When it didn't, they vanished.
I Audited Wildsmith Skin Against AI Shopping Agents. Clinical Trial Data That Machines Cannot Read.
We audited Wildsmith Skin, a British luxury skincare brand born from a 400-acre Georgian estate. They have independent clinical trial data, full INCI lists, and a compelling brand story. Their Shopify descriptions are 7-9 word taglines. AI agents surfaced them on 1 out of 15 queries. The content exists. It is in the wrong place.
The Brand With the Best Product Data Has the Worst AI Visibility
Taylor Stitch has fabric weights, construction specs, and care instructions on every product page. They surfaced on 3 out of 15 AI queries - the worst in our audit set. Good data is necessary but not sufficient.
We Audited Finisterre Against AI Shopping Agents. The Results Surprised Me.
Finisterre surfaced on 9 out of 15 AI queries - the highest in this audit set. Then I looked closer: 6/6 on brand queries, 1/6 on product queries. A brand coasting on reputation while its product data sits empty.
Hiut And Nudie. A Tale Of Two Jeans
We audited Hiut Denim, arguably the best craft denim brand in Britain, against AI shopping agents. They surfaced for their story. They were invisible for their products.
Isabella Vrana Against AI Shopping Agents
We audited Isabella Vrana ,a brand with 175,000 Depop followers, against AI shopping agents. It didn't surface on a single query. Here's the full breakdown.
Lazy Oaf Against AI Shopping Agents. Here's What We Found.
A cult British streetwear brand with over 1 million Instagram followers didn't surface on a single AI query. Not one.
Want results like these?
Get a free AI visibility audit and find out exactly where your products stand.
Get Your Free Audit