Ask an AI assistant where to buy something, and it’s likely to cite a national big box retailer. That’s the finding of a new study from Vaer AI, which analyzed nearly half a million responses from an AI on behalf of Lightspeed Commerce. Search Engine Land’s Danny Goodwin reported the results on Sept. 30, 2026.
Small and local stores may surface as sources in these answers, but the finding highlights the challenge for small and local store owners to get past being cited as a source to actually getting recommended by an AI assistant. The study suggests that the gap between the two is significant.
How The Study Was Run
The research was conducted by Tom Wells, co-founder and principal consultant at Vaer AI, who analyzed 20,000 customer purchasing queries in the United States and Canada. These queries yielded 200,000 responses from artificial intelligence (with the web search feature turned off) and 260,000 responses (with the web search feature on), for a total of approximately 500,000 responses.
The distinction is important because the two modes represent different practical applications; a model that does not use web search relies on its training data, while one that uses it has access to additional real-time information, which should benefit smaller firms that lack the resources of bigger competitors. In reality, the advantage was relatively limited, at around 1.5 percentage points.
One final caveat is that the study was commissioned by Lightspeed Commerce and should be considered as such – a single data point that provides an interesting perspective on the question but does not settle it once and for all.
Big Chains Dominate When Search Is Off
Without web searches, ChatGPT and Google Gemini recommended large national chains as stores 63% to 70% of the time. National chains comprised the overwhelming majority of store recommendations, accounting for nearly two-thirds of all recommendations. Independent or local retailers, meanwhile, were recommended for about 10% of suggestions.
In addition, the researchers tested these models in a head-to-head matchup. When presented with one large retail chain and one small store but with no information about which was which, the models chose the larger one 90% to 94% of the time. Such an outcome indicates that the preference for big-box retailing goes far beyond the mere presence of larger chains in certain product categories.
Live Search Narrows The Gap, But Not By Much
Turning search on did not improve the situation. In all three cases – ChatGPT, Google AI Mode, and Google AI Overviews, big-box retailers continued to dominate as top picks, representing 46–58% of responses.
Another statistic from the study seems particularly telling. In the study, 38% of the retailers mentioned in the responses were small, and 38% were large. However, those big-box stores continued to dominate the top picks, representing 2.5x the share of small retailers.
Stores need to be aware that being mentioned in sources, as found by AI, is not the same as being picked. Goodwin notes that in his “Why We Care” section. It may be easier to find, read, and cite, but it does not mean that it will win out over more established retailers.
Specific Products Make It Worse
Query detail changed the results. More specific shopping questions favored large retailers. For example, when asking questions about buying a gift for a six year old, small stores represented around a third of top recommended retailers. But when the question included a particular product, their share dropped to around a tenth. And over that same shift, the share of picks going to big stores rose from around 40% to 60%.
The study doesn’t say why that might be the case — and it would be speculative to guess. One theory is that the mention of a specific product might cue the model to favor retailers who are more likely to carry it. But again, that’s not something the paper says. What it does show is that small stores perform better when it comes to open ended discovery rather than when it comes to shopping for a specific good.
One Word Changed The Results
The most interesting finding involves a single word. Asking AI for an “independent” store had a dramatic impact on both the recommendations and sources of recommendations.
Adding “independent” increased the share of small or local stores almost two-and-a-half times, from roughly one third of total listings to nearly four fifths, in a randomized sample of the prompts used in the study. On Google’s platforms, the share of sources that were large national chains, fell from around 44% for neutral prompts to as low as 9%.
That’s a significant difference from one prompt to the next. It means that models can highlight small retailers when asked to do so. But they typically don’t include them by default. One caveat is that this jump from around a third to nearly four fifths would be based on a randomized sample of the prompts used in the study (not the full 20,000), so it should be treated more as a suggestion than a definitive finding.
Why “Local” And “Near Me” Didn’t Help
You might expect “local” or “near me” to have similar impact. They did not. The study found that those terms had far less effect than “independent.”
The researchers also suggested a possible reason. An AI may mistake a nearby location of a chain store for a local business. So, if a big retailer has a store around the corner, it technically satisfies a “near me” request, but the chain keeps its dominance in the results.
for a small shop that wants to market itself as local, that’s a concern. The word that shoppers are most likely to use is not the one that gets them results, at least in this data.
What The Study Doesn’t Cover
The study has a few limitations that should be mentioned. First, it only tested the shopping prompts in the United States and Canada, so the results cannot be generalized to other countries or regions. Second, the study only examined the influence of the models on the shoppers’ recommendation and not their actual purchasing behaviors or the likelihood to buy from any of the retail chains after being recommended. Third, the study did not identify why one model preferred one retail chain over the others, and thus the influencing factors are unknown.
However, the limitations of the study do not undermine its overall findings or conclusions. The readers should take the results with caution, keeping in mind that they represent only one study, reflecting the trends in the behavior of the sample in a certain period during an experiment. Individuals considering the findings should conduct further analysis to confirm the results or look for subsequent studies that may provide more insight into the same issue.
What Retailers Can Take From It
The study performed was purely metric-based, not prescriptive – which means anything beyond the figures is speculation. Some few deductions, however, are almost inevitable.
In the context of small-time retail shops, the implications of the study suggest that the mere presence of the shop’s citation may not be sufficient. The figure the study presents appears to lean more towards the importance of whether the shop was mentioned with its position – and for what kind of prompts. Discovery-oriented searches appear to be safer ground for smaller retailers to be suggested than specific product ones.
For marketers targeting independent retail sellers, the implication of the study is the desirability – and perhaps eventual inevitability – of the “independent” label. The figure shows that small shops dramatically benefit from being included in search results with the term “independent” added. Whether consumers will adopt this filtering method is up for debate – but the study does not address it.
For large retail chains, the implication of the study is that the current recommendations systems skew towards favoring them. This trend has the potential to be long-term, but it is not guaranteed – as the algorithms change constantly.
The Bigger Picture
For the first time, this study provides another metric for those tracking the spread of AI: mentions of sources beyond citations. Being cited in the first place represents one category of outcome, while being mentioned among other sources is quite different, as the results show.
Since an increasing number of consumers turn to AI assistants for recommendations and directions to purchase, such responses’ skewing towards national chains could deprive local shops of visits they would otherwise get. This is the kind of risk that Goodwin’s finding touches on without overstating its implications.
The study does not show that the tendency is intentional in any way, merely that it happened across hundreds of thousands of queries; whether the companies behind the assistants change the situation, or local retail finds ways to make it change, should be a topic for further studies.
Frequently Asked Questions
Q1: What did the Vaer AI study find?
AI assistants have a significant bias toward large national retailers over small and local stores when it comes to shopping recommendations. The study used almost half a million AI responses.
Q2: Who ran and commissioned the study?
Tom Wells, co-founder and principal consultant at Vaer AI, conducted it on behalf of Lightspeed Commerce.
Q3: How many prompts and responses were analyzed?
The study includes 20,000 shopping prompts sent to AI across the United States and Canada. The number of responses generated equals 200,000 answers without web search and 260,000 answers with search.
Q4: How often did AI recommend large chains without web search?
When not using a web search, chatbots such as ChatGPT and Google Gemini recommended big-name retailers in 63% to 70% of responses, while independent stores scored only 10%.
Q5: Does turning on web search help small retailers?
Enabling a web search did not significantly impact the outcome. In 46% to 58% of responses across ChatGPT, Google AI Mode, and Google AI Overviews, large national retailers were suggested as the preferred option.
Q6: Are small retailers cited less than large ones?
Both small and large retailers occupied 38% on average of the most mentioned stores. However, the representatives of big-store chains scored 2.5 times higher than those of small-store chains.
Q7: How do specific product searches affect results?
For generic terms, small-store mentions account for approximately one-third of the top recommendations, whereas for specific products, small-store mentions account for only one-tenth. At the same time, representatives of large retailers increased their share from 40% to 60% when the prompt included specific products.
Q8: What happens when a prompt includes the word “independent”?
The number of independent retailers mentioned in a response increased more than two times on average and accounted for nearly 80% of all small stores in a randomized sample. On Google’s platforms, the domination of large retailers decreased from 44% to 9%.




