Atom Foundry Research

2026 AI Recommendation
Readiness Report

Comprehensive analysis of AI recommendation readiness
across thousands of DTC and Shopify stores
atomfoundry.dev · The AI Commerce Graph™
May 2026 · By Daniel Pokorny, Founder
Atom Foundry · atomfoundry.dev · The AI Commerce Graph™
Research

What we found after scanning
thousands of real stores

We run automated scans on Shopify and DTC stores to understand how AI shopping agents actually read, evaluate, and recommend them. The data is live, continuously updated, and ours alone.

—
Stores analyzed
—
Avg AI score
—
AI Invisible Risk
—
Highly Recommendable
Score distribution across scanned stores — May 2026
—
AI Invisible Risk — score below 50
—
Low or moderate confidence — score 50 to 84
—
Reached Highly Recommendable (85+)
AI Invisible Risk (0 to 49) Low confidence (50 to 69) Moderate (70 to 84) Highly Recommendable (85+)
Average AI score by niche — May 2026
The dashed line marks 50, the minimum threshold for Low Recommendation Confidence.
Published Reports

Deep-dive research, fully measured

Standalone reports built entirely from real captured data. No projections, no estimates, every number traceable to a scan or a recorded AI response.

⚡ New · Controlled experiment
Web Search Rewrites 77% of AI Product Recommendations
Same model, same 50 buying prompts. Turn web search on and 77% of recommended brands change. Memory versus retrieval, isolated.
Read the report →
★ Flagship · Cross-category synthesis
The State of AI Recommendations Across Commerce 2026
The full picture. 20,000 recommendations across all five categories, matched to the real stores behind 1,490 brands. One replicated result.
Read the report →
20,000
Recommendations
1,490
Brands
5
Categories
~0
Freq vs Score
Across five categories and 20,000 recommendations, how often AI recommends a brand has no meaningful relationship to how AI-ready its store is. The market measures whether AI sees you. This measures why AI chooses you.
▶ Part Two · The Fame Study
AI Recommends by Fame, Not by Store Quality
The sequel to the cross-category study. We took the 200 most-recommended brands and measured what actually separates the ones AI names from the ones it ignores. The answer is fame, and even fame explains only a quarter of it.
Read the report →
200
Brands
24.9%
By fame
2.1%
By store quality
~12×
Fame over quality
Public fame signals explain twelve times more of how often a brand is recommended than store quality does. The 50 most-recommended and 50 least-recommended brands have nearly identical stores (50.8 vs 49.8). And the recommendations are stable, not random: the same brands win 78 to 91 percent of the time.
▶ Recommendation Intelligence
The State of AI Recommendations in Home & Living
Fifth and final niche. 20 high-intent home prompts, 20 times each. 4,000 recommendations captured, then matched to the real store scores we measure. The most marketplace-driven category yet.
Read the report →
4,000
Recommendations
271
Brands
20
Intents
r = 0.108
Freq vs Score
The fifth category, the same answer. How often AI recommends a brand has no measurable relationship to how AI-ready its store is. IKEA and West Elm are recommended in nearly half of all prompts while their stores score 64 and 60, and the best-built stores barely appear.
▶ Recommendation Intelligence
The State of AI Recommendations in Pets
20 high-intent pet prompts, 20 times each. 4,000 recommendations captured, then matched to the real store scores we measure. Marketplaces separated out.
Read the report →
4,000
Recommendations
405
Brands
20
Intents
r = -0.366
Freq vs Score
The sharpest result of all. Here the link turns faintly negative: the most recommended pet brands tend to have slightly worse stores. Blue Buffalo and Merrick win on fame with stores scoring 30 and 42, while the best-built brands barely appear.
▶ Recommendation Intelligence
The State of AI Recommendations in Coffee
We asked an AI model 20 high-intent coffee prompts, 20 times each. 4,000 recommendations captured, then matched to the real store scores we measure. Marketplaces separated out.
Read the report →
4,000
Recommendations
228
Brands
20
Intents
r = 0.019
Freq vs Score
A third category, the same result. How often AI recommends a brand has no measurable relationship to how AI-ready its store is. Peet's is recommended in 96 percent of prompts and Blue Bottle in 75 percent, while their stores score 56 and 36 out of 100.
▶ Recommendation Intelligence
The State of AI Recommendations in Supplements
We asked an AI model 20 high-intent supplement prompts, 20 times each. 4,000 recommendations captured, then matched to the real store scores we measure. Marketplaces separated out.
Read the report →
4,000
Recommendations
371
Brands
20
Intents
r = -0.015
Freq vs Score
The same result, now in a second category. How often AI recommends a brand has no measurable relationship to how AI-ready its store is. NOW Foods and Nature's Way are recommended in 30 and 14 percent of prompts while their stores score just 14 out of 100.
▶ Recommendation Intelligence
The State of AI Recommendations in Beauty
We asked an AI model 20 high-intent beauty prompts, 20 times each. 4,000 recommendations captured, then matched to the real store scores we measure.
Read the report →
4,000
Recommendations
238
Brands
20
Intents
r = 0.17
Freq vs Score
The central finding: how often AI recommends a brand has no measurable relationship to how AI-ready its store is. Clinique, SkinCeuticals, and Kiehl's are recommended in up to 38 percent of prompts while their stores score just 14 out of 100.
From the founder
D
Daniel, Founder of Atom Foundry
Scanning DTC stores daily since January 2026
What I keep seeing — personal observations
Observation 01
The most surprising thing I keep seeing is that big brands score worse than small niche stores. A well-known fashion brand with millions in revenue can score 12/100. A small skincare startup with 500 Instagram followers scores 68. Brand recognition has zero correlation with AI readiness.
Observation 02
Most founders have no idea this is happening. When I reach out after scanning a store, the response is almost always the same: "We didn't know AI couldn't read our products." The invisibility is silent. There is no error message. The store just gets skipped.
Observation 03
The stores that score highest are not the ones with the best design or the most traffic. They are the ones where someone once added proper schema markup, wrote real product descriptions, and made the return policy readable without JavaScript.
Observation 04
Trust is the hardest signal to fix fast. Missing schema markup can be patched in an afternoon. Low trust scores take longer because trust requires reviews, real policies, contact information, and SSL all working together in a way that AI can parse without rendering JavaScript.
Observation 05
I genuinely believe that the stores optimizing for AI in 2026 will have a compounding advantage through 2027 and beyond. AI agents learn which stores to trust. Early movers build recommendation history. The longer you wait, the harder it gets to catch up.
Key Findings

What the data actually shows

Loading live data...
Not one store out of the first 500 reached Highly Recommendable status (85 or above). The bar for full AI recommendation confidence is high, and no store in our first 500 dataset cleared it.
500 Stores Report · May 2026
Schema markup averages 6.1 out of 20 across the first 500 stores. Most stores have no structured JSON-LD data that AI can reliably parse. Product extraction, pricing, and availability are invisible to AI agents as a result.
500 Stores Report · May 2026
Major retailers scored as low as 4 out of 100. Brand recognition and organic traffic do not translate to AI recommendation readiness. The signals AI needs are completely different from what Google rewards.
500 Stores Report · May 2026
Fashion stores average 41 out of 100 and 52% are AI Invisible Risk. Despite high brand recognition across the category, fashion is one of the weakest verticals for AI recommendation readiness.
500 Stores Report · Fashion Niche · 160 stores
Shopify launched its Agentic Dashboard in May 2026, generating llms.txt, llms-full.txt, and agents.md for every store. The infrastructure is now universal. Content quality is the only remaining differentiator.
Shopify Agentic Analysis · May 2026
Research Topics

What we are studying

These are the areas where we are actively collecting data and publishing findings.

🔍
AI Visibility
How AI shopping agents crawl, parse, and evaluate ecommerce stores. Which signals matter most for getting included in AI-generated recommendations.
Active research
⭐
Recommendation Readiness
What separates stores that get recommended by ChatGPT, Alexa for Shopping, and Perplexity from those that get skipped entirely in buyer queries.
Active research
🛡
Machine-Readable Trust
How reviews, return policies, SSL, and brand signals are interpreted by AI systems, and which formats AI agents can extract without JavaScript rendering.
Active research
🤖
Agentic Commerce
Shopify Agentic Dashboard adoption, llms.txt quality, agents.md checkout flows, and how AI agent infrastructure evolves across the ecommerce ecosystem.
Active research
📊
Ecommerce Benchmarks
Industry benchmarks by niche including fashion, beauty, health, food, and home goods. How does your category compare on AI recommendation readiness?
Coming soon
💰
Commerce Accuracy
How pricing visibility, stock availability, variant structure, and product data quality affect AI confidence when presenting products to buyers.
Coming soon
Our Methodology

How we score stores

We evaluate stores across 8 factors that determine how confidently AI shopping agents can understand, trust, and recommend them.

Read the full methodology
Semantic Visuals & Image Clarity
Alt text quality, image accessibility for AI vision systems. Maximum 15 points.
AI Structured Signals
Schema markup, JSON-LD, llms.txt quality, structured data. Maximum 15 points.
Semantic Clarity & Technical
Content clarity, crawlability, page speed, heading structure. Maximum 15 points.
AI Trust & Transaction Confidence
Reviews, return policy, SSL, trust badges, legal identity. Maximum 15 points.
Commerce & Feed Accuracy
Pricing visibility, stock availability, variant structure. Maximum 15 points.
User Intent Match
Niche positioning, buyer query matching, semantic clarity. Maximum 10 points.
Recommendation Confidence
Social proof, delivery info, guest checkout, return clarity. Maximum 10 points.
External Authority Signals
About page, social links, brand entity, organization schema. Maximum 5 points.
Explore more

Where this connects

The research sits on top of everything else we build. Here is how it links together.

🧮
The Framework
The seven-layer model behind the research: how AI reads, understands, trusts, and recommends a business.
Read the framework →
🔬
Methodology v3.1
The 8-factor, 100-point model used to score every store referenced in these reports.
See how we score →
📊
Store Database
Browse the live AI Commerce Scores behind the brands in these studies. Filter by niche and score.
Browse the data →
📖
Vocabulary
Every term used across the research, defined. Recommendation Share, AI Trust, and the rest.
Open the glossary →
See where your store stands
Free AI Recommendation Score across all 8 factors. Find out exactly what AI agents see when they evaluate your store.
Get My Free AI Score
Free · No signup · Results in 10 seconds