37% of 857 leading Shopify stores can't complete a sale to an AI agent.
We scored 857 leading DTC storefronts with the open Agentic Commerce Score — deterministic, public-signal checks of whether ChatGPT, Google AI Mode, Perplexity and Claude can find, evaluate, and complete a purchase. Only 544 of 857 cleared the agent-buyable bar, and 73% fail a single check: AggregateRating in Product JSON-LD.
Where stores fail, check by check
| Check | Pass | Warn | Fail | n |
|---|---|---|---|---|
| AI crawlers allowed (robots.txt) | 97% | 1% | 2% | 857 |
| Homepage readable without JavaScript | 95% | 2% | 3% | 857 |
| Sitemap discoverable | 98% | 0% | 2% | 857 |
| llms.txt present | 91% | 9% | 0% | 857 |
| Machine-readable product catalog | 91% | 4% | 6% | 857 |
| Required fields (title/image/price/description) | 88% | 9% | 2% | 826 |
| Product JSON-LD (Offer: price/currency/availability) | 78% | 7% | 15% | 806 |
| Brand + product identifier (sku/gtin/mpn) | 47% | 11% | 43% | 806 |
| AggregateRating in Product JSON-LD | 14% | 13% | 73% | 806 |
| Product image alt text | 37% | 31% | 32% | 806 |
| Description depth (≥120 chars) | 80% | 10% | 11% | 826 |
| Agentic-checkout rail (platform prerequisite) | 100% | 0% | 0% | 857 |
| Machine-readable price + availability | 91% | 6% | 3% | 826 |
| Per-variant stock readable | 94% | 4% | 3% | 826 |
| Shipping + returns policies (with terms) | 68% | 27% | 4% | 857 |
Percentages are over the stores where the check could run (n varies because checks that cannot run are n/a, never counted as failures). Grade distribution: A 541 · B 236 · C 27 · D 31 · F 22.
Top 25 by score
| # | Store | Score | Grade | Agent-buyable |
|---|---|---|---|---|
| 1 | beginningboutique.com | 100 | A | ✓ |
| 2 | caddislife.com | 100 | A | ✓ |
| 3 | byoma.com | 100 | A | ✓ |
| 4 | darntough.com | 100 | A | ✓ |
| 5 | dieuxskin.com | 100 | A | ✓ |
| 6 | getquip.com | 100 | A | ✓ |
| 7 | hyperx.com | 100 | A | ✓ |
| 8 | jonesroadbeauty.com | 100 | A | ✓ |
| 9 | littlewordsproject.com | 100 | A | ✓ |
| 10 | mushie.com | 100 | A | ✓ |
| 11 | olaplex.com | 100 | A | ✓ |
| 12 | onesizebeauty.com | 100 | A | ✓ |
| 13 | oofos.com | 100 | A | ✓ |
| 14 | rhodeskin.com | 100 | A | ✓ |
| 15 | sundays-company.com | 100 | A | ✓ |
| 16 | the-citizenry.com | 100 | A | ✓ |
| 17 | thejamesbrand.com | 100 | A | ✓ |
| 18 | thenimetyou.com | 100 | A | ✓ |
| 19 | vegogarden.com | 100 | A | ✓ |
| 20 | 310nutrition.com | 98 | A | ✓ |
| 21 | alaninu.com | 98 | A | ✓ |
| 22 | boxedwaterisbetter.com | 98 | A | ✓ |
| 23 | hyperlitemountaingear.com | 98 | A | ✓ |
| 24 | kravebeauty.com | 98 | A | ✓ |
| 25 | koio.co | 98 | A | ✓ |
Every scanned store has a full breakdown page — browse all 857 stores, or score any store yourself: npx agentic-commerce-score <domain>.
Methodology
Candidate stores were compiled from public “leading DTC / top Shopify” lists (Fastbundle — 200 Best Shopify Stores (2025), Omnisend — 8 per-vertical best-Shopify-stores roundups, PipiAds — Top 100 Shopify Clothing Stores (traffic-ranked), Wisepops — best Shopify stores roundups, Charle Agency — 53 Biggest Brands on Shopify, Eastside Co — 63 Biggest Names on Shopify, TrendTrack — top Shopify stores by revenue, The Hub — 40 Fastest Growing DTC Brands, Skailama — top 25 wellness brands on Shopify, plus a canonical set of widely-known Shopify DTC storefronts, platform-verified during the scan), deduplicated to registrable domains. 955 domains were scanned with the open-source ACS CLI (~8 fixed GET requests each, honest bot user-agent, no JavaScript execution); 955 completed with a score, and the 857 stores reported here are those where a Shopify storefront was detected. Non-Shopify sites and unreachable domains are excluded from headline stats. The scan is deterministic and reproducible: run it yourself — raw NDJSON is committed in the repo's data/ directory.
Known limits: scores are a point-in-time snapshot (2026-07-23); stores change daily. Bot-walled homepages score what a declared agent fetcher actually experiences — which is the point, but means a store behind an aggressive CDN rule can under-score its human experience. One product page is sampled per store for schema checks. Rubric v0.2 thresholds: SCORE.md.