How Often AI Chatbots Misrepresent UK Retailers

Where Data Tells the Story

Home

Support

Creators

Legal

How Often AI Chatbots Misrepresent UK Retailers

Key findings

On average, AI chatbots misrepresent UK high street retailers in 8% of the answers given in response to customer questions on identifying a retail brand's store, location, or contact details.
AI misrepresents electronics and tech shops the most on UK high streets at a rate of 11.9%. A shopper asking about a phone repair shop, for example, is 2.5x more likely to get an inaccurate or missing answer than one asking about a department store.
AI chatbots were most likely to confuse a retailer’s location at a rate of 9.9% of answers, followed by contact details at 9.1% and a brand name or identity was misattributed 5.2% of the time.

Plenty of shoppers are using AI search tools to help them with discovery and purchase decisions. According to a Rithium survey released in March of this year, up to 90% of shoppers already using AI have done so to research products or compare options, and one in five have purchased from a brand they’d never previously heard of, because AI recommended it.

But AI chatbots aren’t perfect in providing accurate information about brands. According to Searchable, the AI visibility platform, their latest research shows 8% of answers that customers ask about high street retailers are inaccurate. That is, they are either incorrect or the AI chatbot isn’t able to answer them, despite verified online sources from the business showing the information exists.

Locations were the most likely piece of information to be misplaced for a high street retailer. Brand names were being misattributed by AI chatbots for the wrong store or entirely different businesses suggested either within the same city or miles outside of it.

Methodology

Searchable queried ChatGPT, Google Gemini and Perplexity with 72,312 questions about 1,328 high street retailers located across 12 major UK cities. Each question was framed as one a prospective customer might ask, which covers a retailer’s identity, location, contact details and the products it sells. Every answer was then scored against verifiable facts for each retailer, and marked wrong only where the evidence clearly supported it. For example, legitimate alternative websites and addresses given without a postcode were credited as correct answers.