Narrative
Walmart.com scored 25 out of 97 and sits squarely in the Invisible band, which is a striking result for the largest retailer in the country. The store publishes a permissive robots.txt and 34 sitemaps, then slams the door shut with PerimeterX at the edge, blocking 59% of fetches and returning a CAPTCHA on the homepage, cart, checkout, returns, and every sampled product page. An agent that follows the welcome mat straight into a verification wall cannot read prices, parse specs, or complete a transaction.
Where the points went
| Problem | Points lost | What an agent hits |
|---|---|---|
| Edge blocking via PerimeterX | 10 | Homepage and product pages return a challenge page, not HTML |
| No Product/Offer schema on sampled pages | 8 | Agent cannot resolve SKU, price, or availability structurally |
| Price absent from raw and rendered HTML | 8 | Nothing to quote, compare, or add to a decision |
| No specs, descriptions, or decision metadata | 11 | Agent cannot answer "is this the right item?" |
| No returns policy, no delivery ETA, no guest checkout | 13 | Every trust and transaction question dead-ends |
The evidence
- PerimeterX blocks 59% of fetches (0/6): Challenge pages logged at
walmart.com/,/cart,/checkout,/returns,/sitemap.xml,/llms.txt, and all 9 sampled product URLs, all from providerperimeterx. Check 4.3 recorded a 90.5% block rate and could not even evaluate rate limiting. - Zero structured product data across the sample (0/6): All 9 sampled URLs returned
hasProduct: false, hasOffer: false, dialects: ["none"]. The sitemap atwalmart.com/sitemap.xmlreportstotal_locs: 0andhas_product_urls: false. - Price is invisible in HTML (0/8):
raw_hits: 0, rendered_hits: 0across all 9 products. Rendered fetches returned status 0, meaning the headless browser was blocked before paint. - Content signals collapse behind the wall (0/11 across 6.1 to 6.4): Classifier notes read "Page content not accessible, robot/human verification barrier prevents data extraction" and "only legal footer text visible." Nothing to summarize.
- Fake urgency flagged (1/2): The phrase "Get as soon as 13 mins" was detected as
fake_urgency, with no legitimate counterpart.
What already works
- Crawler posture is fully permissive: ChatGPT-User, OAI-SearchBot, Claude-User, PerplexityBot, and Amazonbot are all allowed in robots.txt.
llms.txtis published atwalmart.com/llms.txt(though it too returns a PerimeterX challenge).- PayPal is exposed as a machine-native wallet on cart and homepage.
Fix first
- Allowlist verified agent user-agents at the PerimeterX layer for read paths (product, cart, returns, sitemap): recovers ~10 points and unlocks every content check.
- Emit Product plus Offer JSON-LD with price, availability, and shippingDetails on PDPs: recovers ~14 points across 1.2, 2.1, 2.2, 2.3.
- Publish a machine-readable returns policy and delivery ETA copy on PDPs: recovers ~5 points.
The single biggest lever is the edge: every other fix is invisible until agents can actually reach the page.
CDiscoverability
Robots, sitemap, agent endpoints. partial.
EMachine Readability
Schema coverage, raw-HTML price, inventory, variants. broken.
DTransactability
Checkout flow, guest, transparency, payment rails, cart API. patchy.
EAgent Posture
Bot detection, CAPTCHA, rate limiting, stated agent policy. broken.
CTrust Signals
Return policy, merchant verifiability, price history, third-party anchors. partial.
EAgent-Native Content
Specs, descriptions, decision support, delivery, stability, absence of manipulation. broken.