Your Customers Are Sending an AI to Shop for Them
October 7, 2026
By Troy Knott
TL;DR:
Nearly a quarter of online shoppers now start their search inside an AI platform rather than a search bar or a marketplace, and that share climbed from 17% to 24% in ten months. It changes what your product information has to do, because the thing reading it is no longer a person. Here is what the 2026 holiday forecasts actually say, and what is worth fixing before November.
Think about how a customer looks for a plumber, a supplement, or a new office chair this month. Some of them still open a search bar and scroll. A growing share open an assistant, describe the problem in a sentence, and read the answer that comes back. They never see your homepage. They see whatever the model decided to say about you, assembled from your product data, your reviews, your listings, and whatever else it could find and parse.
The Number That Should Get Your Attention
Bain’s 2026 holiday outlook asked online shoppers where they begin. In October 2025, 17% named a generative AI platform. By August 2026, 24% did. Over the same stretch Amazon slipped from 73% to 71%, and retailer and brand websites climbed from 51% to 60%. A further 13% said they plan to use a retailer’s own AI agent, the kind of branded tool Walmart launched as Sparky.
Salesforce measures it from a different angle and arrives somewhere similar. Half of shoppers now report using an AI assistant at some point in the buying journey, a 67% increase year over year, and 74% say they trust product recommendations that come from AI chat. Salesforce expects roughly 20% of holiday ecommerce traffic to originate from AI chat agents.
The two firms measure differently and land on different figures. They agree on direction, which is the part that matters when you are deciding where to spend the weeks before Black Friday.
What an Agent Reads That a Person Doesn’t
A person browsing your site tolerates a great deal. They look at a photo and infer the size. They open a spec sheet in a PDF. They hunt for the shipping policy and find it after a minute of clicking. They fill in the gaps without noticing they are doing it.
An assistant fills in nothing. It works from text it can parse: structured product data, written specifications, review content, stated availability, policies in plain language on the page. A dimension that exists only inside an image is invisible to it. A return policy that lives only in a downloadable PDF may as well not exist. A service area described in a hero graphic is a blank.
This is why businesses with beautiful photography and thin product copy keep finding themselves left out of answers that name weaker competitors. The competitor wrote it down.
Consistency matters as much as completeness. When your hours say one thing on your website, another on your Google Business Profile, and a third on an old directory listing, the assistant has to choose. It frequently chooses wrong, and it does not warn the customer that it was unsure. The customer simply acts on it.
Reviews Stopped Being Decoration
Salesforce found that 41% of shoppers say a brand-owned AI agent makes them much more confident in a purchase, with another 36% somewhat more confident. Retailers that deployed branded shopper agents during the 2025 season saw holiday sales growth of 6.2%, against 3.9% for retailers that didn’t.
Most small and mid-sized businesses will not be deploying an agent of their own this year, and they don’t need to. What they can do is make sure the material an agent draws on is current and specific, because review text has quietly become input rather than ornament.
A dozen five-star ratings with no written content give an assistant almost nothing to work with. A handful of detailed recent reviews that describe an actual situation, a timeline, and an outcome give it language it can use when a customer asks who handles a particular kind of job. Asking customers for a sentence about what they needed, instead of a star, is now a technical optimization as much as a reputation one.
The Season Is Large Enough to Take Seriously
Deloitte projects US holiday retail sales between $1.7 and $1.71 trillion across November through January, with ecommerce growing 7.5% to 8.4% year over year to somewhere between $318.9 and $361.1 billion. Bain forecasts total holiday sales of $1.016 trillion, crossing a trillion dollars for the first time, with non-store sales up 9% against 2.5% growth in physical stores.
A few points of ecommerce growth spread across a season that size represents a great deal of money moving toward whichever businesses happen to be legible to the systems customers are using. Legibility is the operating word. Most of the businesses that lose this year will lose quietly, without a ranking drop or a failed campaign to point at.
It is also worth noting where the season is not going. Salesforce found that 77% of shoppers still prefer physical stores for holiday shopping, and 42% visit a store after researching online. The AI conversation happens upstream of that visit rather than instead of it, which is good news for businesses with a physical location and bad news for any of them treating their online product information as optional.
What to Fix Before November
Start with your product and service data. Every item needs written specifications in text on the page, not in an image and not in an attachment. Sizes, materials, compatibility, turnaround times, service areas, what is included and what isn’t. If a customer would ask it, write it.
Then reconcile your listings. Name, address, phone, hours, and service description should match across your site, your Google Business Profile, and any directory that still ranks for your name. Kill the listings you no longer control if you can, and correct the ones you can.
Then make your policies plain. Shipping cutoffs, return windows, warranty terms, and booking lead times belong in readable text on a page an assistant can reach without a login.
Then ask the assistants directly. Open ChatGPT, Gemini, and Perplexity and ask the question a customer in your category would ask, without naming your business. Read what comes back. It is the closest thing to a free audit you will get this quarter, and it tends to be uncomfortable in a useful way.
Every year there is one shift that businesses notice a season too late. This is that one. If a customer asks an assistant for a recommendation in your category and your business never comes up, you won’t find that in a January ranking report. It shows up as revenue that quietly went somewhere else.
— Troy Knott, Spring Digital
Find Out How AI Sees Your Business
If you are not sure how your business shows up when someone asks an AI assistant about your category, that is worth checking before the holidays arrive. Spring Digital’s free site and SEO analysis covers visibility, structure, and technical performance across your digital presence.

Founder and CEO of Spring Digital, Troy Knott brings over 15 years of experience in web strategy, SEO, and digital marketing leadership. He’s passionate about guiding teams and clients through scalable growth, combining sharp business insight with a deep understanding of digital ecosystems.


