Searchable, the AI visibility platform, has released new research revealing how inaccurately the UK’s three most-used AI chatbots answer everyday questions about retailers on British high streets.
After prompting ChatGPT, Google’s Gemini, and Perplexity with more than 72,000 questions and grading every answer against each retailer’s verified information, the test results show roughly two in three (64%) businesses had at least one false fact returned about them.
The AI chatbots returned an answer to a question 98% of the time for queries relating to a retailer’s brand identity, contact details, and location. However, one in 16 answers provided by the AI engines were false.
The most common false answer misplaced high street retailers into the wrong postcode, at a rate of one in 10. This occurred despite the test prompts giving clear identifiers on the town or city the business was located in and asking the tools to specify the locations of all relevant addresses.
In 15% of cases where the wrong location was given, AI chatbots returned an address of a retailer that was 20 miles away or more. The median distance between a shop’s correct location and the incorrect location given by an AI chatbot was over one kilometre.
AI chatbots also recommended the wrong website: around one in 15 of the chatbots’ website answers pointed to a dead link, a lookalike, or a different business entirely.
The tests also show AI chatbots attributed the wrong brand to a shop’s name, at a rate of one in 40 answers.
Searchable co-founder Chris Donnelly explains why this is a challenge for Britain’s high streets to overcome in reaching new and existing customers: “Overall, our research into this issue has shown that AI inaccuracies are more likely to impact smaller businesses than larger brands across multiple sectors. This can be partly explained by big companies having a larger pool of third-party sources online that provide accurate information about them.
“For a smaller bricks-and-mortar retailer, if their online visibility mostly centres around its website and a Google Business listing, that’s a relatively thin trail of information for AI systems to learn from and to represent in their answers.”
Searchable’s analysis also shows that the levels of misinformation were unevenly distributed across the three Large Language Models (LLMs) tested, with Perplexity returning inaccurate answers in 10% of cases, compared to 4% for ChatGPT and 5% for Gemini.
According to a 2026 Rithium survey of UK-based and US-based online shoppers who already actively use AI products, 90% use LLMs for product research and comparison, and 53% use them to choose a retailer. Inaccurate AI answers risk leading customers astray from high street businesses when these tools are seen as trusted partners in the purchase journey, with 58% of those surveyed also saying they lose trust in a brand when AI provides the wrong product information.
Donnelly says that high street retailers can take relatively small steps today to reduce the likelihood of AI systems misrepresenting them: “Ensure your website and third-party business directories are up to date and clearly state your company’s essential services. Beyond that, consider how local press mentions, trade association listings and customer reviews can help you show up more positively and correctly online. These are all factors that influence AI’s interpretation of a brand.”
