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

Lululemon scores 25 out of 100 and lands in the Invisible tier, which is the polite way of saying that to an AI shopping agent, the world's most valuable athleisure brand effectively does not exist. The irony is thick: the company welcomes every major crawler at the robots.txt layer (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, all green-lit) and then hands them a website engineered to defeat them. The doors are open. The building is empty.

The single largest wound is price invisibility in raw HTML, an 8-point check on which Lululemon scored zero. Across ten sampled product pages, prices appeared in the raw response exactly zero times, and even after full JavaScript rendering only 2 of 10 pages surfaced pricing DOM (Pace Breaker Short and ABC Classic Fit Short). The rest returned network errors. For an agent trying to answer "how much does the ABC jogger cost," Lululemon is a wall of Lit web components (<product-widget-element>, lit$888034805$) that never resolve to a number. Compounding this, structured product data coverage (check 1.2, worth 6 points) is also a shutout: not one of the ten sampled product URLs exposed Product or Offer schema. Zero. On a site where every product page presumably has a price and a SKU in the backend, this is a choice, not an accident.

Transactability compounds the discoverability problem into a full collapse. Neither /cart nor /checkout returned usable content to the scanner, which zeroes out checkout flow structure (6 points), guest checkout (4 points), price transparency (4 points), and cart API accessibility (3 points). That is 17 points that never had a chance. Combined with the absence of any agentic payment rails detected on product pages, the entire transactability dimension effectively signals to an agent: complete this purchase somewhere else. Given that 30 of 32 scanner fetches returned network errors while the two that succeeded came back cleanly (no 403s, no CAPTCHAs, no rate limits), the culprit reads less like hostile bot defense and more like a client-side rendering architecture that simply refuses to serve HTML to non-browser clients.

Trust signals and agent-native content round out the damage. There is no reachable /returns page, no footer or about content exposing merchant registration details, and zero third-party review aggregator presence detected across Trustpilot, Bazaarvoice, Yotpo, Google Shopping, and eight other platforms. Lululemon has millions of reviews somewhere; none of them are legible to an agent evaluating merchant credibility. Meanwhile the two product pages that did render revealed that the underlying copy is actually excellent (detailed fabric composition, fit guidance, True Fit widget data, model measurements), but because that content sits behind the same rendering layer that hides the prices, specification completeness, conversational descriptions, and decision-support metadata all score zero out of a combined 9 points.

The good news, such as it is: crawler posture is perfect, bot detection is not aggressive, and there are no dark patterns. The infrastructure is welcoming; the payload is empty.

Lululemon's single highest-leverage move is server-side rendering (or a parallel static HTML surface) for product pages that exposes price, availability, and Product/Offer JSON-LD, which alone would unlock roughly 20 points and move the brand from invisible to legible overnight.