Invisible. Agents can't meaningfully engage with this storefront.

Patagonia scores 32 out of 100 and lands in the Invisible band, which is a remarkable result for a brand of this stature. The site does the polite things: robots.txt welcomes GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, anthropic-ai, Applebot-Extended, Bytespider, and Amazonbot without a single disallow, and both /.well-known/ai-plugin.json and /llms.txt are live. But hospitality is not the same as legibility, and once a crawler arrives, there is essentially nothing machine-readable to consume. Compounding this, the entire transactional surface was serving a maintenance interstitial during the scan ("Hang Tight! Routing to checkout..."), which means agents attempting to actually transact hit a wall.

The single largest point loss is price in raw HTML, worth 8 points and scoring zero. Across all ten sampled URLs, including /shop/mens/tops/t-shirts and /shop/gear/bags/backpacks/laptop, neither the raw HTML nor the rendered DOM contained a detectable price. Four of those URLs returned a rendered status of 0, suggesting the client-side rendering pipeline is failing outright for agent-shaped traffic. If a shopping agent cannot extract a price, the product does not exist in its consideration set. This is the most expensive check on the report, and Patagonia scored nothing.

Structured product data is the second wound: zero out of six on check 1.2, zero out of six on 2.2 (product attributes), and zero out of five on 2.3 (inventory signals). Not one of the ten sampled pages emitted Product schema, Offer schema, or any labeled attribute pair. Combined with a sitemap at https://patagonia.com/sitemap.xml that returns 200 but contains zero locs and no lastmod (another 4 points forfeited on check 1.3), the discoverability story is: crawlers are invited in and then handed an empty room. Schema.org Product with Offer, AvailabilityShipping, and priceValidUntil on PDPs would flip roughly 20 points on its own.

Transactability collapsed largely because of the maintenance state, but the underlying architecture is also weak. Guest checkout could not be evaluated (0 of 4), price transparency signals for shipping and tax were absent pre-auth (0 of 4), no machine-native payment rails were detected across six product pages plus cart, checkout, and homepage (1 of 3), and no cart REST endpoints or form POSTs were exposed (0 of 3). Even discounting the maintenance interstitial, the absence of any Shop Pay, Link, or comparable rail signal on a catalog this large is a deliberate architectural choice that shuts out agentic checkout.

Content-layer signals rounded out the damage. Specification completeness, conversational descriptions, decision-support metadata, delivery date transparency, and price stability all scored zero, for 13 points lost in the agent-native content dimension alone. The classifier saw "low" signal on every sampled URL. Trust signals were nearly as thin: no third-party review platform detected among Trustpilot, Yotpo, Bazaarvoice, or the eight others checked, and no return policy content was retrievable because /returns was also serving the maintenance page.

The single highest-leverage move is to ship Product plus Offer JSON-LD with price, availability, and shippingDetails on every PDP and category page, which would unlock roughly a quarter of the total score before touching anything else.