Send an AI agent on a test purchase. See which step it reaches, what stops it and what to fix.
You’re checking your own store. But you already know every step. The agent starts from zero.
An example on a demo store: from the page, through the code layers, to checkout.
Every test is a session recording, a step timeline and the exact stopping point. We keep observations apart from guesses.
After clicking “Add to cart”, the message “Select a size” appeared. The agent went back to the size list twice and ended the test without adding the product.
The size buttons are images with no text and no accessible labels. The agent most likely did not recognise them as options.
Add visible text or an aria-label to each size, e.g. “Size 42”, and mark the selected variant with aria-pressed.
Type your store’s domain. We start with a quick scan of public pages and show a preliminary result.
After you leave an email, the agent walks the purchase path: product, variant, cart, payment.
You get the run, the stopping point and guidance on what to fix.
Monitoring repeats the test and alerts you when a change to your store makes the result worse.
The Agent-Readiness Score combines the quick scan (40%) and the agent run (60%). It describes how ready the purchase path is for the CanItBuy test agent. It is not a sales forecast.
The free test shows whether and where an AI shopper drops off. A paid plan proves the purchase works after your fix, and keeps checking every week.
| Free$0one-off | Most popular to startMonitor$29/ month | Pro$69/ month | |
|---|---|---|---|
| Technical scan (40 pts) | ✓ | ✓ | ✓ |
| Agent in a real browser, live | 1 test | weekly | daily |
| Report with the stopping point | ✓ | ✓ | ✓ |
| Session recording with agent view | excerpt | full | full |
| Re-test after your fix | – | ✓ | ✓ |
| Alert when the score drops | – | ✓ | ✓ |
| Score history: before and after | – | ✓ | ✓ |
| Fix guides for your platform | – | – | ✓ |
| AI models tested | 1 | 1 | 3: Claude, GPT, Gemini |
| Check my store | Choose Monitor | Choose Pro |
You see a recording, not a vague grade. You know exactly what to fix.
A new theme or plugin can quietly break the cart. Monitor catches it.
Agent-ready stores collect the first orders from a new channel.
An agency? The Agency plan ($199/mo): many stores and reports with your logo. Waitlist open.
Card payments via Stripe. Prices exclude VAT. Plan checkout opens soon.
No. The agent stops at the payment step. It never enters personal or payment data and never places an order.
No. The test shows how far the CanItBuy test agent got. Other agents, including consumer ChatGPT, may behave differently. The result does not guarantee recommendations or sales.
No. The test uses public pages. Monitoring starts after you confirm the domain is yours.
A 0–100 score: 40% from the quick scan, 60% from the agent run. We show the full score only after both steps.
We say so plainly and give the reason we could establish, e.g. the site not responding or blocking the test browser.