How do AI shopping agents handle discount codes and promotions?
Agents find codes the way determined shoppers do: they read the banners, test the fields, and stack what stacks. Stores need a policy on agent-applied discounts before the agent checks out, not after.
Agents are thorough coupon hunters
An AI shopping agent tasked with buying something does not just click add to cart. It reads the announcement bar, scans the homepage for promo codes, checks the cart for a discount field, and tries the codes it finds. Some agents go further, searching the web for codes for your store the way a bargain-hunting shopper would. Whatever your promotion setup allows a persistent human to do, an agent will do faster and more systematically.
This is not a hypothetical. As agent-driven shopping grows, the discount field becomes an API of sorts: a predictable input that a machine will probe. Stores should decide deliberately what agents may and may not do with promotions, because the default is that they will do everything the field permits.
The stacking question comes first
The highest-risk behavior is stacking: an agent applying a welcome discount, a sale price, a free-shipping threshold, and a loyalty reward in one order, in combinations no human would think to try. Agents do not feel awkward about it. They optimize. If your promotion rules technically permit a stack, expect an agent to find it.
Audit your promotion rules for stackability before agents do it for you. Decide which combinations are intended and configure the rest out. This is good hygiene regardless of agents; agents just shorten the time between a misconfiguration and its exploitation from months to days.
Public codes vs targeted codes
Not all codes deserve the same exposure. A sitewide sale code in the announcement bar is public by design; an agent using it is behaving like any shopper. A single-use influencer code, a win-back code emailed to one customer, or an internal testing code is not public, and an agent applying it is using something it was never offered.
The practical defense is code hygiene: single-use codes that actually expire after one use, targeted codes tied to the customer account rather than the code string, and test codes that never work on the live store. Agents cannot abuse what does not work. The stores that get surprised are the ones where a "private" code has worked for anyone who typed it for two years.
Decide the agent promotion policy
Write down what your store intends: agents may use publicly advertised promotions; agents may not use single-use or targeted codes; loyalty pricing requires the account holder. Then enforce it technically where you can (code scoping, account binding) and state it in the terms of service where you cannot. The policy matters for two reasons: it guides your configuration, and it gives you a basis for canceling orders that clearly abused non-public codes.
Consider what the agent experience should be for legitimate promotions. An agent that cannot tell whether a code applied correctly may retry, abandon, or misreport the price to its user. Clear cart messaging about which discounts applied, and why a code was rejected, helps agents complete purchases cleanly instead of thrashing.
Watch the logs for agent coupon probing
Coupon probing has a signature: repeated cart submissions with different code values, often in quick succession, sometimes from the same session that browsed normally first. Distinguish this from a human trying two codes; the volume and speed give it away. Log discount attempts with the code value, the session, and the outcome, and review the pattern weekly as agent traffic grows.
Rate-limit the discount field per session. A human tries a handful of codes; an agent can try hundreds. A modest limit (a dozen attempts per session, say) inconveniences no real shopper and blunts automated probing. It is the same principle as login rate limiting, applied to the promo field.
The promotion strategy angle
There is an upside worth naming. Agents that reliably apply public promotions complete more purchases, and completed purchases are the point. A store with clean, well-structured promotions (clear eligibility, honest stacking rules, codes that work as advertised) converts agent traffic better than a store with a tangled promo setup that agents misread. In the agent era, promotion clarity is a conversion feature, not just a marketing detail.