Amazon is actively promoting a feature called Auto Buy to Prime members right now.

The pitch to consumers: set a price target on a product, and Rufus watches it every 30 minutes. When the price hits your number, it buys automatically — saved card, saved address, done. No human required at the purchase step.

Amazon is marketing this with a specific stat: Auto Buy users are saving 20% per purchase on average. That’s in the consumer marketing. Amazon wants shoppers to know this feature pays off.

Twenty percent. Think about what that means on your SKUs.

Here’s how the mechanic works. A Prime member looks at your $32 protein powder. They want to pay $25. They set $25 as their Auto Buy target. That request sits active for up to six months. They move on with their life. Meanwhile, Rufus is pinging your listing twice an hour.

You run a spring sale. Price drops to $24.99. Auto Buy fires. Rufus buys the protein powder at $24.99 without the consumer doing anything at all. They don’t see your coupon. They don’t engage with your storefront. They don’t read your new A+ content. The agent just executes, at the price they decided they were willing to pay months ago.

That’s one transaction. But the feature allows up to 200 simultaneous requests per account. And Amazon’s Big Spring Sale is running through March 31 — the first major sale event with Auto Buy live.

So here’s what’s actually happening: every time you run a promotion, you’re telling consumers what your floor is. They set Auto Buy targets at or below that floor. Those targets accumulate — across your whole customer base — and they sit there, waiting. The moment you discount again, the purchases fire. At the bottom of your range. Every time.

This isn’t a theory about what might happen. The 20% savings figure is Amazon’s own marketing. That savings has to come from somewhere. It’s coming from the spread between a brand’s regular price and the lowest price they’ve trained the market to expect.

Frequent promotions have always trained price-sensitive consumers to wait for sales. That’s not new. What’s new is that the waiting is now automated and patient in a way humans aren’t. A person who decides to “wait for a deal” gets impatient. They buy at full price eventually, or they forget. An agent doesn’t forget. It just pings your listing every 30 minutes for six months.

The compounding effect is real. A consumer who buys from you twice a year at $32 is worth $64. A consumer who has Auto Buy set at $25 and gets triggered twice is worth $50. Same person. Same product. Different infrastructure. You just built the infrastructure that extracted $14 from your margin.

Rufus was active in 38% of Amazon shopping sessions during Black Friday. Auto Buy is the monetization layer on top of that. The question isn’t whether this feature will matter — it’s already live, already being promoted, already scaling. The question is whether your pricing strategy accounts for it.

A few things worth doing now:

  1. Look at your promotion history over the last 12 months. The lowest price you’ve run publicly is probably the price point where thousands of Auto Buy targets are sitting. That’s your de facto floor, whether you intended it or not.
  2. Think hard about promotional frequency. A brand that discounts six times a year is a brand whose agents have learned to wait. A brand that discounts once — predictably, seasonally — has a floor that’s harder to game because there’s only one window.
  3. If you’re going to discount, consider mechanisms that don’t set a permanent price signal. Subscribe & Save. Bundle deals. Coupon codes that aren’t visible to price-tracking tools. The goal is to move units without broadcasting “this is what we’re willing to take.”

Auto Buy isn’t the enemy. Thoughtless discounting is. Auto Buy just made the consequences faster and more systematic.

Your promotional calendar is now also your pricing floor. Act like it.