Retailers are working to revamp their digital storefronts and make them stand out to AI chatbots as more and more people use the likes of ChatGPT to find the best Black Friday deals and holiday gifts.

The rise of so-called agentic commerce, where people use chatbots to either research or outright purchase new products, has persuaded many retailers to revamp both the code on their websites and the way they write about the products they’re selling in the hopes of grabbing the attention of an AI crawler.

“You need to be able to customize your product data to be optimized for those channels,” said John Johnston, senior director of online experiences at Richmond, B.C.-based Trail Appliances. 

The appliances retailer is already getting a decent amount of traffic to its website via chatbots and is currently redesigning it in the hopes of bringing in even more. AI-driven traffic to its site is still relatively low, but Johnston expects it to rise sharply in the coming months. “It’s going to take another year for the general public to really warm up to these tools,” he said.

For that to happen, the technology likely needs to improve. Chatbots are prone to ignoring a shopper’s budget or struggling to find the newest version of a product. It’s something the companies building these models are working on. Google tries to build trust by linking to source information for the answers its AI serves up, said Eric Morris, managing director of retail for Google Canada.

Still, retailers see opportunity. During its most recent earnings call in November, Shopify said its merchants have seen a roughly seven-fold increase in AI-driven traffic this year and an 11-fold increase in orders stemming from AI searches. In a recent survey, Google found 43 per cent of Canadian shoppers plan to chat with AI about their holiday shopping.

The actual use of AI for shopping is likely even higher, according to Morris. People doing a regular Google search, for example, will almost always see an AI overview before any standard search results. That’s caused a steep drop-off in web traffic as people are seemingly less likely to click through when AI provides a summary right at the top of the Google search result page.

It helps that shopping through chatbots is getting easier. In late September, OpenAI updated ChatGPT to let people research and buy things through its chatbot. Makeup brand Glossier, shapewear maker Spanx and Kim Kardashian’s Skims were the first Shopify merchants to make their products available for purchase directly in ChatGPT.

Johnston said Trail Appliances has found that people shopping using AI tend to have bigger budgets, a trend the company is watching closely as it works to revamp its website for AI crawlers. To do that, his company is working with Merchkit, which helps retailers retool their websites for the demands of AI. 

While people have long asked Google more general questions, AI chatbots, due to their conversational nature, tend to make people be way more specific. Before AI, someone might have searched Google for the best hiking boots. In the AI era, they might ask for the best waterproof hiking boots to wear in winter in Vancouver. Have an answer to that on your website and there’s a good chance a chatbot will find it and regurgitate it to someone shopping, said Bijan Vaez, Merchkit’s co-founder and CEO.

Answering those super-specific queries, though, involves writing—and lots of it. “There’s all this net new content that will actually help you get discovered on ChatGPT that I don’t think had really been considered before,” said Vaez. There’s a risk, though, that some companies use AI to create content for their websites that appeals to AI crawlers. These crawlers will, in turn, grab this AI content and regurgitate it to people using chatbots to help them with their holiday shopping. 

Sponsored content that’s scooped up by AI crawlers may also get tangled up with genuine results, making it difficult to distinguish between paid posts and organic recommendations. Google’s Morris said its AI systems don’t necessarily prioritize content created by AI—though he added that people will soon start to see some sponsored content in its AI overviews.

Trail Appliances is writing text for its new website to try and anticipate the questions people may ask a chatbot, said Johnston. What’s the best dishwasher under $1,000 that has a hidden handle? What’s the best laundry machine if you’ve got pets? What’s the cheapest five-element convection oven with a good warranty? 

Beyond answers to specific questions, chatbots are also hunting for a lot of detail. Two-sentence product descriptions no longer cut it, said Vaez. Instead, retailers should be adding detailed FAQs sections for each product they sell. It’s a great spot to hide a lot of information, while also answering a lot of the questions that might get picked up by an AI.

Morris claimed Google prioritizes “rich, original content” and detailed descriptions with, for example, size guides or care instructions for apparel. Photos and videos can also help an AI better understand a product. Amazon’s Rufus, another AI shopping assistant, searches photos to see how the product is being used, said Vaez, so it can build up more knowledge about it.

Retailers that hope to win the race for chatbot shoppers will also need to work on the backend of their websites, Vaez said. That means ripping out JavaScript, which is slow to load and can be missed by AI scrapers, and making sure HTML is clean and accessible.

Strong customer reviews are also important, Vaez said, as ChatGPT and other AI agents look for “social proof” as to whether a product or idea is good or not. That’s likely to put more pressure on retailers to try and persuade people to write reviews of products they buy. Or, in some cases, to pay people to write fake reviews in an attempt to game the system.