Twenty years of conversion tactics don't work on AI shoppers
Conversion-rate optimization was built to exploit one thing: human System-1 bias. That exploit is closing.
Countdown timers, “only 2 left,” strike-through pricing — all engineered for a brain that fears loss and rushes decisions.
New HBR research ran thousands of simulated purchases across four leading models and four product categories. The persuasion playbook didn't just stop working on AI shopping agents — the more advanced the model, the more it treated overt urgency as something to discount. Only two signals moved agents reliably in the expected direction: star ratings up, price down. Everything else was noise.
The strategic read for CMOs: as agent traffic grows, large parts of your CRO stack become dead weight, and the manipulation tactics that once lifted human conversion may actively suppress agent selection.
The moat shifts to fundamentals you can't fake — authentic reviews, honest pricing, machine-readable product data. And each model behaves like its own segment, with its own quirks.
Worth reading before you greenlight your next urgency experiment.