Study: AI Shopping Picks Big Retailers 9 Times in 10

AI shopping assistants picked the bigger retailer 90 to 94 percent of the time in head-to-head tests, according to a study Lightspeed released Sept. 29.

If you run a high-ticket Shopify store, you are the smaller retailer in that matchup. Shoppers are asking ChatGPT and Google’s AI which store to buy from, and the first data set large enough to measure it says small stores lose most of those calls before the shopper ever sees a search result. The gap was widest on product-specific queries, which is exactly how high-ticket buyers search. I teach that model at Ecommerce Paradise, so I read this one closely.

Below is what the study measured, where it is weak, and the five things I’d test this week. If you are new to the model, start with my guide to what high-ticket dropshipping is, because the advice here assumes you sell big-ticket items through a niche store.

Work on discovery without watching your fixed costs creep up. Northwest Registered Agent keeps your renewal price the same as year one with no upsells, so the compliance line on your P&L stays boring while you go fix the part of the business that moves revenue. See Northwest Registered Agent →

Lightspeed Study: ChatGPT and Google AI Chose Big Retailers 90-94%

Lightspeed Commerce, the retail and hospitality software company, commissioned the research from Vaer AI, according to Lightspeed’s release. Tom Wells of Vaer AI ran it. He called it “one of the first large-scale attempts to measure how generative AI influences retail discovery.”

Search Engine Land reported on Sept. 30 that the study covered 20,000 shopping prompts across the U.S. and Canada. That produced 200,000 AI responses without web search and 260,000 with search enabled. Lightspeed says the prompts spanned 10 product categories in four cities, Los Angeles, San Francisco, New York and Montreal, and tested ChatGPT, Google AI Mode and Google AI Overviews, per 6ix Retail’s summary.

Without web search, ChatGPT and Google’s Gemini recommended major retailers 63 to 70 percent of the time and independent stores about 10 percent of the time, according to Search Engine Land. When the models were shown one large and one smaller retailer with no size cues, they chose the larger one 90 to 94 percent of the time.

Turning live search on narrowed the gap only a little. National chains still made up 46 to 58 percent of recommendations with search enabled, per Lightspeed. Google AI Overviews recommended no local store at all in 68 percent of shopping responses.

Specificity made it worse. Local retailers appeared among the top recommendations in roughly one-third of broad searches but only about one in ten for specific product queries, per Lightspeed’s figures as summarized by 6ix Retail.

The one lever the study found was a single word. Adding “independent” to a prompt more than doubled small-store recommendations, lifting their share from roughly one-third to nearly four-fifths, according to Search Engine Land. Dax Dasilva, Lightspeed’s founder and CEO, put it this way: it “cut large-chain recommendations from about 44% to as little as 9%.” Terms like “local” and “near me” barely moved the result.

A companion Censuswide survey of 2,000 North American consumers found that 56 percent have used AI for shopping decisions and 50 percent would shop locally more if AI made independent retailers easier to find, according to Roastbrief’s summary. Lightspeed says only 33 percent of respondents think AI should prioritize small businesses, against 13 percent who favor large brands.

Dasilva said “visibility within those recommendations will become a critical driver of growth for retailers.” The timing matters for you. Adobe projects 2026 online holiday sales of $275.1 billion, per Digital Commerce 360. I covered the furniture line of that forecast, up 7.3 percent, in my Adobe holiday breakdown.

How AI Assistants Became a Shopping Channel for Store Owners

This study did not land in a vacuum. OpenAI has been building a paid layer on top of ChatGPT all year. I reported the early numbers, a 0.73 percent U.S. click-through rate with ecommerce sellers mostly absent, in my ChatGPT ads write-up. On Sept. 30, Search Engine Land reported that OpenAI expanded ChatGPT Ads with bulk product campaigns and deeper reporting.

The checkout plumbing is moving too. Shopify now lets AI agents submit orders on your store, which I broke down in my Shopify agents post. Meta’s Muse assistant plugs into Shopify and Stripe, covered in my Meta Muse report.

Ads and checkout are the two ends of the pipe. The Lightspeed study measures the middle, which store the assistant names when a shopper asks what to buy. Nobody had put a number on that before this week.

Several things limit how far you can push these findings. Lightspeed sells software to independent retailers and paid for the work, so the framing favors its customers. The “small retailers” in the test were local independents in four cities, not online-only stores, so this is not a direct read on a dropshipping business. Coverage also does not report whether any recommendation turned into a sale. The models were tested on what they recommend, not on what shoppers buy.

The sources disagree slightly on scale. Search Engine Land’s figures add up to about 460,000 responses, 6ix Retail says roughly 460,000, and Roastbrief says nearly 500,000. I am using the 460,000 figure. Counterpoint from a skeptical reader is easy to imagine: large retailers have more pages, more reviews and more links, so any model trained on the web would name them first. The study does not test that explanation, and nothing in the coverage rules it out.

What Big-Retailer AI Bias Means for a High-Ticket Shopify Store

My read is that the one-in-ten number for specific product searches is the figure to stare at. A buyer shopping a $2,000 outdoor kitchen or a $4,000 mobility scooter does not type “furniture store near me.” They type a brand, a model and a question about delivery. That is the query type where small stores almost disappeared.

My guess at the cause, and it is only a guess, is that a model asked for a safe answer names the retailer it has seen most often. The “independent” result suggests the models can surface small stores when the prompt asks for them. The default prompt just does not ask. You cannot change how shoppers phrase things, so the work is making your store the obvious answer when the prompt is specific.

Here is the hypothetical math, with numbers I made up to size the problem. Say 1,000 specific-product AI searches a month happen in your niche. If independents collectively get about 10 percent of recommendations, that is 100 slots shared among every small store. Win half of them and you have 50 slots. At a 20 percent click rate that is 10 visits, and at a 2 percent conversion rate that is 0.2 orders, or roughly $360 at a $1,800 average order. Today this channel is small, and nothing in the study says otherwise.

That is why I would not rebuild a store around this. I would also not ignore it. The cost of a test is a few hours, and the Censuswide numbers say more than half of shoppers already use AI to decide. A channel that is 1 percent of your orders now can be 5 percent by next holiday season.

Do not cut your paid spend because of this study. The research measured organic AI answers, not Google Shopping ads, which are paid placements. Google is already blurring paid and free listings, which I covered in my sponsored grid report. My view is that a store with a weak conversion system gets hurt by AI discovery and by ads, which is why I wrote why high-ticket Google Ads need a complete conversion system.

Here are the scenarios I’d use. If more than 5 percent of your orders say they came from an AI answer, put real time and budget into this channel. That threshold is mine, not the study’s. If it is between 1 and 5 percent, rerun the test monthly and fix the cheap things. If it is under 1 percent, finish the holiday push first and revisit in January.

Trust signals are where a small high-ticket store can compete. You win against Home Depot-sized catalogs by being the authorized dealer who answers the phone, quotes freight accurately and stands behind the warranty. The study did not test whether those signals change recommendations. I would still build them, because they convert on your own site whether or not an assistant ever names you.

Authorized dealer status starts with the supplier. A directory like Inventory Source can help you find US-based suppliers that will put dealer terms in writing. Wholesale2B is another place to look when you want wider coverage, though I would still get warranty and return terms from the manufacturer directly.

Reviews are the other lever. Tools like Yotpo collect and display them on product pages, and I would put the effort into your top 10 SKUs before anything else. For tracking, Ahrefs has a Brand Radar feature built to track brand mentions in AI answers. I have not run it on a high-ticket store, so check the plan limits before you buy.

Owned audience matters more now, not less. If a discovery layer you do not control starts favoring bigger sellers, your email list is the part nobody can reroute. Klaviyo is the email tool I point store owners to, and a repeat buyer who asks for you by name skips the assistant entirely.

If that list is more work than you want to run alongside the store itself, that is what my team does through the turnkey done-for-you service, where we build and launch the store with these trust signals built in.

Want my team to scale the store you already have while AI shopping rules settle? See the scaling service →

Test Your Store in ChatGPT and Google AI Mode This Week

Five actions, in this order:

Build a 40-prompt test sheet. Take your top 10 SKUs and write four prompts for each: the exact model name, “best [category] under [your price point],” “where to buy [model],” and the same question with the word “independent” added. Run them in ChatGPT and in Google AI Mode, and log whether your store is named, whether it is linked, and who beat you. Run the same sheet again in 30 days.
Add a “how did you find us” question to your post-purchase survey with these options: Google search, Google ads, ChatGPT, Google AI answer, a friend, other. Count the answers for 30 days. This is the number that decides which of the scenarios above you are in, and my guide to what your ecommerce numbers are really telling you covers how to read it.
Put your product page next to the biggest competitor’s page for the same model. Compare authorized dealer status, warranty terms, freight cost and time, phone number, and spec detail. Fix the gaps on your top 10 SKUs first, using the product template in your Shopify theme to keep the layout consistent. My Yotpo setup guide for high-ticket stores walks through getting real reviews on those pages.
Keep Google Shopping running and tighten it. The study did not measure paid placements, so treat them as your reliable channel while AI results settle. Start with how to turn Google Shopping clicks into sales for high-ticket products.
Pick a tracker and a keyword tool. Set up the Ahrefs Brand Radar feature or a manual monthly log, and check the keyword side with SEMRush. If you want a second set of eyes on your results, book a call at my discovery page.

If you want to cut the manual work, I rounded up the best AI SEO automation tools for ecommerce stores. Start with the test sheet before you buy anything.

Frequently Asked Questions

Does this mean ChatGPT is biased against small stores?
The study shows a pattern in what the models recommend, not why. Adding the word “independent” changed the results sharply, so the models can surface small stores when asked. The cause is untested.

Should I add the word “independent” to my site copy?
The study changed the prompt, not the page. I’d add it where it is true, like an About page or a dealer statement, and test the result. Do not expect it to work on its own.

How reliable is a study Lightspeed paid for?
Treat it as a strong signal with a sponsor. Lightspeed sells to independent retailers, the sample was local stores in four cities, and the coverage reports no sales data. I would run my own test sheet before acting on the headline.

Does this apply to a dropshipping store?
The direction probably does, but the size of the effect is unknown. The “small retailers” in the study were local independents, not online-only stores. Your own test sheet is the only way to know how your niche behaves.

Which niches face the least big-retailer competition?
Niches where the product is specialized, the supplier relationships are hard to get, and the big chains do not stock the full range. My free niches list is a good place to start. My high-ticket niches guide covers the products worth selling.

How many shoppers use AI to decide what to buy?
In the Censuswide survey for Lightspeed, 56 percent of 2,000 North American consumers said they have used AI for shopping decisions. Using AI is not the same as buying from it, so check your own order data.

Want 1-on-1 help fixing your store’s trust signals before holiday traffic hits? Get the coaching details →

Run the test sheet this week and write down what you find, even if the first answer is that nobody names you. That is a baseline, and you can only improve a number you have. Subscribe to the YouTube channel for daily breakdowns. More breaking news coming through the day.

If this was useful, these go deeper:

Trevor Fenner

Trevor Fenner is an ecommerce entrepreneur and the founder of Ecommerce Paradise, a platform focused on helping entrepreneurs build and scale profitable high-ticket ecommerce and dropshipping businesses. With over a decade of hands-on experience, Trevor specializes in high-ticket dropshipping strategy, niche and product selection, supplier recruiting and onboarding, Google & Bing Shopping ads, ecommerce SEO, and systems-driven automation and scaling. Through Ecommerce Paradise, he provides free education via in-depth guides like How to Start High-Ticket Dropshipping, advanced training through the High-Ticket Dropshipping Masterclass, and fully done-for-you turnkey ecommerce services for entrepreneurs who want a faster, more hands-off path to growth. Trevor is known for emphasizing sustainable, real-world ecommerce models over hype-driven tactics, helping store owners build scalable, sellable, and location-independent brands.