Test ecommerce website: the walk a buyer takes

You open the shop from your own bookmark, signed in, and go straight to the product you added last week. The menus work because you built them. A stranger arrives from a search result on an older product page, types a word your product name does not use, and lands on a category that lists half of what you published. Your view can include drafts a buyer never gets. A test that starts where a buyer starts is the only way to see the shop they actually meet.

• The category page lists the same products a signed-out visitor can open, with no dead links.
• A search for the plural or the other common word for a product returns that product.
• A product with options cannot be added to the cart with an option that is out of stock.
• The cart count in the header matches the cart page after the first item is added.
• A product page opened directly from a link still shows a way back to the shop and to the cart.

Signed in, you see drafts, preview pages and products you have not yet published to the public catalog. A stranger sees none of that. The gap shows up as a category that lists fewer items than your admin says it holds, or as a link from a newsletter that opens a product nobody can buy. The tell is a count. Open a private window, load one category, and count what a signed-out visitor sees. Then count the same category in your admin. If the two numbers differ, the difference is the list of pages a stranger will hit and you will not.

Cached pages add a second gap. A page loaded earlier in your browser can show an old price or an item that sold out, while a new visitor gets the current figure, or the other way round. Walk the shop in a private window, not in the browser you use to manage it, so the page you read is the one a buyer reads.

Search is the first place a buyer tests your words against yours. A product called a soy candle does not turn up for a buyer who types candles, if the search only matches the exact title. A category name that uses a word buyers do not use is the same problem one level up. Product options add a step. A size or color that is out of stock can still be selectable, and the failure shows only when the buyer presses add to cart, which is the moment they decide whether to trust the shop.

The cart is the last place a stranger checks whether they did something. An item added, a count in the header that stays at zero, and a cart page that is empty all tell the buyer that the shop lost their order. The screenshot shows the header and the cart side by side, so you can see whether the count and the page disagree.

Many buyers never start at your home page. They arrive on a product or a category from a search result or a shared link, and the first screen they see is one you did not design as a front door. Start a run from a product URL, not the home page, and see whether that page gives a way to the rest of the shop and to the cart. A product page with no header, or a header with no cart, is a dead end for anyone who did not come in through the front.

A simulated visitor can click through the menu, type in search, choose options and open the cart. It reaches the payment step and stops there, because it cannot pay with a real card. The walk therefore covers everything up to the card, and the card step is one you check yourself with an order you place and refund. Run it as a guest, since visitors cannot sign in with Google or single sign on.

Checkout flow testing

Catalogs change constantly: a new collection, a retired product, a sale that changes the price on a category page. Each change can leave a link that points at a product a stranger cannot open. Rerun the same walk after every catalog change, on a preview address, and compare the signed-out count with the number you see as the owner. A GitHub Action can run the walk on each preview deploy.