Attention online retailers! Bot shoppers are about to change everything

Autonomous bots acting as virtual shoppers for consumers and businesses are likely to be common by within the next two to three years — but online retailers are nowhere near ready for what’s coming.

The arrival of bot shoppers is going to force a change in how retail sales are conducted at all stages of the buying process, from how products are displayed, promoted, priced and sold. And even when online sites are optimized for all those digital shoppers, nagging issues are almost certain to arise.

Think of it this way: given that bot shoppers will likely represent a large number of clients during a single  shopping “trip,” how would returns work? Many retailers today limit how many returns any one shopper can make to try and thwart a type of fraud called “renting” or “wardrobing.” 

If, for example, a store limit is 20 returns within six months, what happens when a bot represents 18,000 shoppers? For that matter, what about those loyalty points often used to encourage sales and bring back customers? Will the bot keep those points (meaning they really go to the bot’s owner, such as OpenAI, Microsoft, AWS, Google or Anthropic)? Or do they accrue to human clients in proportion to the purchases made? Or somewhere else?

For some of these bot-forced issues, there’s no clearcut answer. Bots are software, their clients are human. Each will be shopping with a different set of parameters. There are some shopping elements bots don’t care about at all, such as beautiful high-resolution photographs and dramatic marketing copy extolling the elevated lifestyle indicated by any particular purchase.

But human clients might really care about those elements. Does the bot ignore those elements when shopping, and then download them after a purchase to share with the human client?

How does the autonomous shopper bot even work? Does it actually make a purchase on its own and simply alert the  client, “Got the shoes and the refrigerator you wanted. They both should arrive to you on Tuesday”? Or does it report back with a small list of recommended purchases, letting the human decide on (and complete) the purchase? 

Here’s a key e-commerce question underlying these scenarios: Will retailers encourage bots to visit their regular websites and hope for the best? Or will retailers need to create special bot-friendly versions of their online stores, streamlined to deliver maximum efficiency to AI agents? Does that lead to true bot-to-bot interactions — with no humans invited?

Working for me? Or thee?

In the old days, travel agents used to be free  they made their money by being paid by various hotels, restaurants, airlines and car rental businesses. That often left the traveler wondering whether the travel agent pushed a particular hotel or travel package because it’s the best value for the money — or because they were getting a kickback?

That gets us to the most challenging issue with shopping bots: Trust. Will human shoppers wonder whether the bot is making purchases for the shopper’s benefit or the bot owner’s benefit? 

How will payment card fraud be handled when a bot is involved? What if an attacker tricks the bot into making fraudulent purchases? Will merchants and banks reimburse for that routinely?

In the earliest days of e-commerce, Visa and other card brands created a program called Zero Liability, which still exists today. The idea was to give people a reason to try online retail sites. Would something similar work for bot shopping?

Human browsing, machine parsing

“I think this really calls into question the underlying processes for retail today, (because) we are used to doing everything through a human lens,” said Frank Diana, managing partner at Tata Consultancy Services. “Retailers will need to optimize product data not just for human browsing, but for machine parsing — clear, structured, and API-accessible information becomes essential. This will “force retailers to rethink fixed pricing, loyalty models, and discount structures.”

Retailers need to reimagine just about every process, said Julie Geller, a principal research director at Info-Tech Research Group.

“It’s not just about visibility,” she said. “It’s about whether your product data, pricing and availability are structured in a way these systems can understand and act on. The real shift is from persuading humans to enabling machines. The retailers who design for that will win.”

Geller makes a good point, but doesn’t go far enough. Most bots will easily be able to interpret the English that retailers post for human shoppers. But that’s not the most efficient and effective way for them to get information because the information isn’t optimized for them. And that difference will prove critical for a couple of reasons. 

First, the more efficient a site is for bots, the more likely it is they will visit often. That will ultimately translate to revenue, which is why creating a bot-friendly — or even a bot-only — site makes sense. There’s also the efficiency issue. If a site is designed for maximum bot efficiency, it could process a visit faster. When a retailer is calculating revenue per-hour, especially during the holidays or during a special sale, that speed could make a big difference to the bottom line.

Long-time retail analyst Leslie Hand, who tracked retail IT issues for IDC for 16 years until January, agreed that retail IT executives have a lot to do to prepare for bot shoppers, which she estimated would start to meaningfully arrive — likely hitting 1% of all retail revenue — by 2028.

“They need to buil …agentic capabilities and integrations,” Hand said. “The biggest change for retailers is allowing automated decision processes to happen outside their four walls. They will need to give up control to agents to make contextualized offers not of the brands design, but of course within the parameters established. Bots will need to have access to data about customers, products, and inventory [and] fulfillment.”

There is a potential positive side effect from retailers prepping for the bot onslaught — the bot-em line, dare I say: they will likely be forced to dramatically increase the SKU-level details for all products. That will also make shopping a lot easier for human shoppers.