Nearly 60% of Americans use GenAI for Online Shopping

Nearly 60% of Americans use GenAI for Online Shopping

The survey finds that 59% use generative AI for shopping, with 25% preferring ChatGPT over Google for product research despite privacy concerns.

A new Omnisend survey of 1,200 American consumers reveals that generative AI is reshaping online shopping habits in major ways. Fifty-nine percent of respondents already use Gen AI tools for various shopping tasks, and far fewer feel reluctant to let AI handle transactions – 32%  compared to 66% back in February

However, despite widespread adoption, 85% report lingering concerns over privacy,  personalization, and overall AI fatigue.  

“Openness to AI is clearly on the rise, and businesses can no longer ignore it,” says  Marty Bauer, E-commerce Expert at Omnisend. “But there’s a big difference between offering suggestions and inserting AI into every interaction in a way that feels intrusive. As privacy remains a major concern, ecommerce businesses must be  crystal clear about where, why, and how they use AI.”  

65% opt for ChatGPT, 25% say it outperforms Google  

With nearly two-thirds of Americans using Gen AI tools as their online shopping  assistants, 29% say it makes the experience less overwhelming – and they’re  turning to it most often for:  

  • Product research – 57% 
  • Personalized recommendations – 45% ● Finding deals – 40% 

While various tools like Perplexity AI and Amazon Rufus were mentioned, 65% of  those using Gen AI when shopping online prefer ChatGPT. In fact, one-in-four agree  its product recommendations are better than Google’s. 

“Googling often means ads, SEO content, and dozens of open tabs before you find  what you need,” explains Bauer. “In contrast, generative AI tools tend to give you a  distilled answer similar to that of a knowledgeable friend, making it easy to see why  shoppers would choose one over the other to help with purchase decisions.”  

85% remain unconvinced as others warm to AI handling checkout  

In just five months, consumer acceptance of AI completing purchases on their behalf  has nearly doubled – a significant shift as OpenAI reportedly develops checkout  features for ChatGPT similar to those already offered by Perplexity AI, highlighting  how close AI-driven commerce is to becoming mainstream.  

But while openness to AI might be growing, many shoppers remain cautious. In fact,  85% report having concerns, including:  

  • Privacy and data security – 43%  
  • AI misinterpreting preferences – 37% 
  • Irrelevant recommendations – 35% 
  • AI being overused in ecommerce – 26% 

“A drop in hesitation shows that convenience can quickly outweigh fear when the  experience is smooth and transparent,” Bauer notes. “But the fact that so many  shoppers still cite privacy and accuracy concerns means trust will define AI’s next  chapter in ecommerce.” 

Tips for ecommerce brands leveraging generative AI  

As companies like Amazon, Walmart, and Mastercard advance agentic AI tools – not  to mention automated checkout features reportedly coming to ChatGPT – lingering  scepticism shows that retailers must build trust before expanding AI’s role in the  purchase journey. Bauer recommends the following:  

  1. Giving shoppers control is non-negotiable. Be transparent about how AI  works, what data it uses, and allow instant opt‑in/out. The more control  people have, the more trust they’ll place in your tools.  
  2. Design for permission. Build in approval points before AI executes any action,  such as adding items to a cart or initiating checkout. Having the option to  approve or decline reassures customers at best and prevents  misunderstandings at worst.  
  3. Keep a visible human touch. Make it easy to switch from AI support to a live  person – especially for high‑value or personal purchases where people tend to  trust people more.  

Methodology  

The survey was commissioned by Omnisend and conducted by Cint in July 2025. A total of  1,224 U.S. consumers were surveyed. Quotas were placed on age, gender, and place of  residence to achieve a nationally representative sample among users.