Where AI Belongs in E-Commerce (and Where it Doesn’t)

The early data on artificial intelligence shopping is in and it has a story to tell.

Shopify reported that shoppers who arrive at e-commerce sites through AI platforms like ChatGPT, Perplexity, and Gemini are 50 percent more likely to convert to purchase than those coming from organic search. Not only are they more likely to make a purchase but when they do, they spend 14 percent more per order. More than half of these AI-referred sessions land directly on a product page, ready to buy.

It's clear that AI is becoming one of the most effective discovery channels for brands, however, AI discovery doesn’t tell the whole story for consumers.

The same consumers who are happy to let AI help them shop want very little to do with AI after they buy. A survey from earlier this year found that 79 percent of Americans strongly prefer interacting with a human over an AI agent, and 56 percent have negative feelings about companies that lean on AI in their customer experience.

So, what are retailers supposed to do? On one hand, you have a signal pointing you toward more AI, while on the other hand, you have a signal pointing you toward more humans. The answer is actually pretty simple once you look at it through the lens of the consumer.

Consumers Love AI When it Helps Them Buy

Shopping with AI can feel useful, especially for consumers who are saturated with options and information. You describe exactly what you need, the AI narrows it down and you click through to a product that matches. All that time you would typically spend on research, comparison and selection gets compressed into a single conversation and a few minutes.

This is also why AI-referred traffic converts so well. The consumer has already done the heavy lifting with AI before they ever reach your site.

Consumers Still Want Humans When Something Goes Wrong

We’ve all experienced the emotional impact of a stressful customer experience. There are certain moments when a person just wants to feel heard. Maybe it’s a billing issue or a delayed order. Other times it’s a damaged package or the wrong size. Customers want to connect with someone who can read the situation and make a judgment call.

An AI agent that insistently delivers the wrong answer in that moment creates unnecessary stress and degrades the trust your brand has spent years building.

The Hybrid AI Customer Support Model

There is a better way: an e-commerce customer support platform that uses both AI and human customer service. This hybrid AI customer support leverages the best of both approaches.

AI handles the quick wins like order status, tracking, return policies, FAQs, ticket triage and routing. These are the queries where speed is the priority and customers prefer a fast self-service answer over waiting in a queue.

Humans take the harder stuff like complaints, escalations, refund decisions, sensitive conversations, and anything that requires the ability to read between the lines. This is where human empathy, judgment and accountability earn the kind of loyalty that lasts for years.

How does it look in practice? Brands use AI to make their human employees faster and better informed, while agents see the customer’s full history, AI-generated context and suggested responses. Both can improve the overall customer experience.

What to Plan For

Here’s what you need to know as you plan to weave AI into your e-commerce business:

1. Lean into AI at the start of the journey when shoppers are choosing what to buy.
2. Keep humans at the forefront of the experience post-purchase, where every interaction is a chance to build or break trust.
3. Audit your support flow by asking: Where are we sending customers to a bot when they really need a person? Where could AI be handling a question that doesn’t need a human to be answered?

Customers have spoken and they've told us how they want to be treated. It’s time to adjust your customer support to meet their needs.

Sachin Jaiswal is a serial entrepreneur and technology leader, and the co-founder and CEO of Kim, an AI-native customer support platform.