Narrative
Allbirds scores 61 out of 100, a mixed signal that masks a strange bifurcation. The site is genuinely welcoming to AI crawlers at the door and technically competent underneath, with full-credit machine-readable pricing, Shop Pay and Amazon Pay rails, and a Shopify cart endpoint any agent can hit. But the moment an agent tries to actually transact or verify what it is looking at, the site starts throwing up walls: CAPTCHAs on the cart, no stated agent policy, no third-party reviews, and product data that thins out on exactly the pages where shoppers land from a search engine.
The most expensive miss is CAPTCHA prevalence, which scored zero out of four. Both reCAPTCHA and hCaptcha were detected on the cart and on the homepage, and the check flagged this as transaction-blocking. Agent-friendly crawler headers in robots.txt mean nothing if the same site fires a challenge at any automated session trying to add to cart. Allbirds has told GPTBot, ClaudeBot, PerplexityBot, and Amazonbot that they are welcome, then bolted the door behind them. Pick a lane.
The second big drag is checkout flow structure, at two out of six. The classifier could not evaluate the checkout page at all because it requires an initialized cart, and a 403 was returned at https://allbirds.com/checkout. This is standard Shopify behavior, but combined with the CAPTCHA finding it means an agent has essentially no way to reason about the purchase path before committing. Guest checkout availability took a related two-out-of-four hit for the same reason: the classifier had low confidence because there was nothing to classify.
Third, structured product data coverage came in at three out of six. Seven of nine sampled products carry proper Product plus Offer JSON-LD, but two do not: the womens-tree-pipers-luna and womens-tree-runner-go-blizzard-hanami-night pages return no structured product markup at all. These are live PDPs on a Shopify store, which makes the gap look like a template regression rather than a design choice. The same two URLs also came back as low-signal on specification completeness (2/4) and conversational descriptions (1/2), suggesting a specific product template is silently degrading agent readability across multiple dimensions.
Fourth, trust signals leak points in two places. Third-party trust anchors scored zero out of two: no Trustpilot, no Yotpo, no Judge.me, no aggregate rating hits anywhere in the checked platforms. For a brand with Allbirds' consumer recognition, the absence of any machine-verifiable review corpus is a self-inflicted wound. An agent choosing between Allbirds and a competitor with a Yotpo widget has no third-party signal to weight the decision. Merchant verifiability adds another two lost points because the footer stops at "Allbirds, Inc." with no physical address or registration ID.
Finally, stated agent policy scored zero out of two, with no terms or privacy content available for the classifier to evaluate. Combined with a fully permissive robots.txt, this leaves agents guessing about what use is actually sanctioned.
The biggest opportunity is removing the CAPTCHA and 403 barriers on cart and checkout so that the strong pricing, payment rail, and cart API foundations Allbirds already has can actually be used by the agents it is inviting in.
BDiscoverability
Robots, sitemap, agent endpoints. mostly strong.
BMachine Readability
Schema coverage, raw-HTML price, inventory, variants. mostly strong.
BTransactability
Checkout flow, guest, transparency, payment rails, cart API. mostly strong.
DAgent Posture
Bot detection, CAPTCHA, rate limiting, stated agent policy. gappy.
CTrust Signals
Return policy, merchant verifiability, price history, third-party anchors. partial.
DAgent-Native Content
Specs, descriptions, decision support, delivery, stability, absence of manipulation. gappy.