The Agentic Commerce Readiness Report · live data · updated 2026-08-12

69% of top DTC stores fail the robot customer.

69% of 100 well-known DTC stores failed when an AI shopping agent tried to add a product to the cart (Coalbird Agent Audit, 2026-08). Every number on this page comes from real agent runs — 581 errands and counting, re-run on a rotating schedule. No surveys, no self-reporting: a browser-driving agent either finished the errand, provably, or it didn't.

  • stores audited100
  • failed add-to-cart69%
  • returns policy unfindable46%
  • cleared every errand22%
  • fought a popup40%

Add-to-cart is where agentic commerce breaks

Agents found the product on 80% of stores — discovery mostly works. Completion doesn't: only 31% of stores let an agent get an item provably into the cart. The gap between "the agent found your product" and "the agent could buy it" is the agentic-commerce readiness gap, and it is where stores silently lose agent-assisted sales today.

Readiness by category

Among categories with five or more audited stores, home handles the robot customer best (57% add-to-cart failure) and food handles it worst (100%).

Category Stores Add-to-cart failure Returns unfindable Cleared all errands
apparel 26 69% 27% 6 / 26 details →
beauty 17 65% 82% 1 / 17 details →
home 14 57% 14% 5 / 14 details →
footwear 11 64% 45% 3 / 11 details →
accessories 10 80% 60% 2 / 10 details →
kitchen 8 63% 50% 3 / 8 details →
food 7 100% 71% 0 / 7 details →
outdoor 3 67% 0% 1 / 3 n too small
pets 2 100% 100% 0 / 2 n too small
kids 1 0% 0% 1 / 1 n too small
eyewear 1 100% 100% 0 / 1 n too small

Methodology

Each store gets the same three errands a human shopper runs without thinking: find a named product and report its price, find the returns policy, and put the product in the cart. A real browser-driving AI agent performs each errand in one visit with a hard step budget. Grading is strict and binary — an errand counts as passed only when the outcome is provable (a price reported, a policy located, an item verifiably in the cart). The bird never checks out, never creates accounts, never enters personal data. Stores are re-audited on a rotating schedule; this page always reflects the latest merged data.

Citing this report

"69% of 100 well-known DTC stores failed when an AI shopping agent tried to add a product to the cart (Coalbird Agent Audit, 2026-08)." — please link to coalbird.co/report. Per-store evidence lives in The Hundred; each store page shows the step-by-step trace.

Your store's number is knowable.

Start with a scoped trial: baseline missions, screenshots, failure traces, a rerun, and a readout.

Start a trial for your store