How do I prevent bots from buying my products?

Combine four things. Choose a release format that does not reward speed, such as a draw. Enforce purchase limits across accounts by device, payment method and address. Detect automation on the product page, the cart and the checkout API. Review orders after the sale and cancel linked ones before shipping.

The release format sets the terms of the contest. A first-come sale at an announced time rewards whoever submits the fastest request, and software is faster than people. A draw assigns the right to buy before any checkout takes place, so speed has less value. A queue with random positions sits between the two. For the most sought-after releases, a draw changes the advantage of speed, although duplicate entries still need controls.

Purchase limits are where most shopping bots concentrate their effort. They are built to run many accounts, each with its own email address, card and slightly varied delivery address. A limit counted per account does nothing against this. Count it across device, payment method fingerprint and normalised address, and apply it on the server at checkout.

Detection has to cover the routes bots use. Many do not load the product page at all. They call the add-to-cart and checkout APIs directly, having worked out the product and variant identifiers in advance. Protect those endpoints, and check that requests to them come from a session with a plausible history. Keep identifiers for unreleased products out of public pages, feeds and sitemaps until the release.

Account preparation helps. Requiring an account created and verified before the release date, with a confirmed email address and phone number, raises the cost of each bot account. Favouring accounts with a purchase history for early access rewards real customers.

Review after the sale catches what passed. Before shipping, group orders by delivery address, card fingerprint and device. Cancel groups that exceed the limit, and refund them. This step is unglamorous and effective, because the attacker loses the stock and has funds tied up in refunds.

Tell customers how the release works. Publish the format, the limit and the fact that linked orders are cancelled. Clear rules reduce complaints and give support staff an answer.

Expect adaptation. Operators test your defences on ordinary products before a release, and change tools afterwards. Review each release: what share of traffic was automated, how many orders were cancelled as linked, and how quickly the product appeared on resale sites. Those three numbers show whether the next release needs a different format.

How Kairal helps

Kairal Bot Protection checks requests on the pages and APIs connected to your integration. You can configure protection per endpoint and review recorded decisions in Events; your application continues to enforce purchase limits and authorisation.

Kairal Bot Protection