AI Search Tools Flunk Local Business Accuracy Tests, Threatening Customer Trust
A recent vendor-run test has exposed a significant vulnerability in the burgeoning landscape of artificial intelligence-powered search, revealing that AI tools returned at least one false fact about 64% of UK high street retailers. This startling inaccuracy rate, shared with Retail Focus by vendor Searchable, highlights a critical blind spot for many marketing teams and raises concerns about the reliability of AI-generated information for consumers seeking local businesses. The most prevalent error identified was the misattribution of business locations, with AI systems frequently placing establishments in incorrect postcodes.
This pervasive inaccuracy underscores a challenge that has largely escaped the tracking efforts of marketing departments. When AI systems like ChatGPT or Google’s AI Mode field questions about a specific business location, there is currently no inherent mechanism to ascertain the veracity of the information being disseminated. Businesses are left in the dark about whether their premises are being accurately represented, potentially leading to confusion or lost custom if the AI confidently asserts that a business is closed, lists services it doesn’t offer, or provides an erroneous address. While previous coverage has explored strategies for achieving visibility within AI-driven search, including insights from Dan Taylor on prompt tracking, this analysis delves into the critical, yet often overlooked, aspect of what AI is already communicating about businesses and how to proactively identify and rectify these inaccuracies before they impact customer perception.
The shift from traditional search engine results pages (SERPs) to AI-synthesized answers represents a fundamental change in how consumers discover and interact with local businesses. In the past, local search provided a multitude of comparable data points for consumers to evaluate, such as map packs, customer reviews, official business websites, and competitor listings. This allowed users to weigh various factors before making a decision. AI search, however, streamlines this process into a single, comprehensive narrative that describes a business, often before a user even navigates to their website or business profile.
This new paradigm of being "described" rather than simply "ranked" introduces a different set of challenges. While rankings can fluctuate based on numerous algorithmic factors, AI-generated descriptions can present information with an unwarranted air of authority, even when inaccurate. Furthermore, the visibility of these AI-generated answers varies significantly depending on the nature of the search query. Whitespark’s analysis indicated that AI Overviews appear in approximately 15% of direct local-intent searches, such as queries for specific professional services within a given city. However, for informational questions, AI Overviews surface in a striking 92% of cases, and for hybrid queries, which often involve decision-making processes like weighing the necessity of hiring a professional, they appear in 97% of instances. This contrasts with traditional local packs, which were present in 93% of direct local-intent searches. The data suggests that AI Overviews are becoming increasingly dominant for more nuanced, informational, and decision-oriented queries.
Adding to these findings, SOCi’s Local Visibility Index, which scrutinized over 350,000 business locations, found that ChatGPT recommended only 1.2% of them. This figure is notably lower than the appearance rates for the same brands in Google’s traditional local 3-pack, according to SOCi’s own metrics. The growing reliance on AI for local business recommendations is further evidenced by BrightLocal’s Local Consumer Review Survey. This report revealed a dramatic surge, with 45% of consumers now utilizing ChatGPT or similar AI tools for local business recommendations, a substantial increase from just 6% the previous year. This rapid adoption highlights the urgency for businesses to address the accuracy of AI-generated information.
Unpacking the Inaccuracies: A Deeper Look at AI’s Missteps
Searchable’s recent testing, designed to quantify the extent of AI misinformation, yielded concerning results. In an initial test involving 165 London-based businesses, Searchable subjected ChatGPT, Gemini, and Perplexity to 13,365 questions covering services, contact details, business size, and founding dates. The AI-generated answers were then rigorously cross-referenced against official Companies House records and other authoritative business profiles.
According to reports by CMOTech, an astonishing 93% of these London businesses had at least one fundamental piece of information presented incorrectly or omitted entirely by the AI tools. The impact of these inaccuracies appeared to be disproportionately felt by smaller businesses, with half of them receiving at least one false fact, compared to 32% of larger enterprises. A subsequent, broader test focused on UK high street retailers, involving over 72,000 queries. Retail Focus reported that approximately one in every sixteen answers provided was incorrect. Notably, errors related to incorrect postcodes occurred at a significant rate of one in ten, even when the search prompts explicitly specified the town or city.
Chris Donnelly, co-founder of Searchable, articulated the underlying issue for smaller bricks-and-mortar retailers: "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." This observation points to the data-starved nature of smaller businesses’ online presence as a potential driver of AI inaccuracies. The categories of errors identified in these tests mirror the complaints frequently voiced by business owners in Google’s support forums. Users report instances where incorrect details are presented with unwavering confidence, with some lamenting the detrimental impact these AI-driven falsehoods are having on their businesses.
The Evolving Landscape of Local Search Visibility
The transition from traditional search engine optimization (SEO) to AI-driven discovery marks a profound shift in how businesses need to approach their online presence. Traditional local search historically provided a tangible, trackable digital footprint. Metrics such as impressions and clicks were meticulously monitored in tools like Google Search Console. Ranking fluctuations were a constant concern, often triggering immediate investigations into potential causes. However, the current AI search ecosystem lacks a comparable system of alerts for businesses when an AI answer incorrectly states operating hours or erroneously declares a location as closed.
Traditional search, while not always alerting businesses to their precise representation, at least directed customers to sources they could independently verify. These included a business’s own listing, its website, and reviews with associated dates. This allowed for a degree of consumer-driven fact-checking. AI, conversely, consolidates these disparate sources into a singular, synthesized textual answer. Customers often perceive this synthesized answer as factual without the opportunity to cross-reference it with the original sources. When an AI description is inaccurate, there is no visible ranking position to scrutinize or traffic dip to analyze, as the error is embedded within text that the business owner may never have seen.
Major consumer-facing AI platforms generally do not offer businesses proactive alerts at the location level when their information is misrepresented. Google’s own documentation acknowledges that AI responses may contain errors, and individual AI Overviews typically include a feedback link. While this feedback mechanism can be used to report and correct inaccuracies, it functions as a reactive measure rather than a proactive monitoring system. Its efficacy hinges entirely on a consumer or business owner first noticing the error and then taking the initiative to report it.
Navigating the Multi-Platform AI Ecosystem
The complexity of testing AI accuracy is further compounded by the fact that different AI search platforms operate with distinct methodologies. AI Overviews are now integrated into standard Google search results, offering a conversational experience directly within Google Search. Google’s Gemini functions as a separate AI assistant, while ChatGPT and Perplexity are distinct products developed by OpenAI and Perplexity AI, respectively.
Each of these platforms processes, selects, and synthesizes information through its own unique algorithms. Consequently, a single query can elicit markedly different answers depending on the platform used. This variability is evident in Searchable’s retail data: Perplexity was found to generate inaccurate answers in 10% of cases, compared to 5% for Gemini and 4% for ChatGPT. This disparity means that a location that appears accurately in Google’s AI Overviews might still be mischaracterized on ChatGPT. Therefore, testing a single AI system provides an incomplete picture, necessitating a comprehensive approach across multiple platforms.
A Proactive Strategy for Auditing AI-Generated Business Information
To effectively address the inaccuracies in AI-generated information, businesses must adopt a systematic and proactive auditing process. This begins with anticipating the types of questions potential customers are likely to ask. Common inquiries revolve around operating hours, available services, location suitability, and specific offerings. Compiling these questions into a standardized list ensures that each business location is evaluated consistently.
The next step involves running these standardized questions through the prominent AI search platforms, including Google’s AI Overviews and AI Mode, as well as ChatGPT, Gemini, and Perplexity. Where possible, tests should be conducted without the influence of saved conversation history or personalized search settings to ensure objectivity. Meticulous record-keeping of both the prompts used and the resulting answers is crucial for analysis.
Given that AI responses can exhibit variability even when presented with the same prompt, it is advisable to repeat queries multiple times. The gathered information should then be organized into distinct categories, such as factual errors, missing information, or perception issues. It is important to differentiate between straightforward factual errors and omissions, which can often be corrected, and reputational issues like sentiment or recommendation order, which require a different strategic approach. Particular attention should be paid to rectifying mistakes that could directly deter potential customers, such as incorrect operating hours or the erroneous assertion that a location is closed.
When AI answers cite specific sources, it is imperative to verify the accuracy of the information within those cited sources. Businesses should focus on correcting any inaccuracies within information they directly control and, where necessary, formally request corrections from third-party platforms. Regular, ongoing checks are essential, as correcting a source does not automatically update all AI-generated answers. Employing a consistent question list and a detailed log will streamline this process, especially for businesses with multiple locations. The frequency of these audits should be adjusted based on the number of locations managed and the typical rate at which business details change.
Rectifying AI Misinformation: A Multi-Faceted Approach
The initial discovery of an AI-generated inaccuracy is only the first step in a multi-stage resolution process. Effectively correcting these errors requires a careful adjustment of the underlying data sources that AI systems rely upon, followed by rigorous re-evaluation to determine if the answers have improved. Each iterative adjustment brings businesses closer to achieving more accurate and dependable AI-generated information.
If an AI answer inaccurately reflects a business’s operating hours, for example, the immediate priority is to ensure consistency across all owned digital assets. This includes verifying that the Google Business Profile, the company website, and individual location pages all present accurate and concordant information. Conflicting details across these platforms are frequently the root cause of AI misinterpretations. The fundamental principle of Name, Address, Phone number (NAP) consistency, a long-standing tenet of local SEO, remains paramount. This becomes even more complex and critical when managing multiple business locations.
Following the correction of information on owned platforms, attention must turn to third-party sources that businesses do not directly control. When an AI answer cites a specific external source, it is essential to examine that source, ascertain its publication date, and review the actual information it provides. The AI may be drawing data from an outdated directory listing, a customer review, or even a page detailing a business with a similar name. Rectifying these external inaccuracies often necessitates direct outreach and communication with the platform hosting the erroneous information.
This is where the disparity noted by Chris Donnelly between smaller and larger businesses becomes particularly relevant. The higher error rates observed in smaller businesses can be attributed, in part, to a more limited online presence, providing AI systems with less comprehensive data to process. Research on factors influencing ChatGPT citations supports this conclusion: a more robust and consistent digital footprint across various online platforms provides AI models with richer, more reliable material for generating accurate business descriptions.
If the primary issue is not accuracy but rather the lack of recommendations or mentions by AI tools, businesses need to shift their focus to visibility strategies. Experts like Dan Taylor have provided valuable insights into tracking how AI systems represent brands over time and understanding how personalized AI search evolves with individual user behavior. These areas are crucial for addressing the broader question of how businesses can effectively "win" in this evolving AI-driven discovery landscape.
The Horizon of AI Monitoring Tools
The development of tools specifically designed to monitor AI-generated content for businesses is still in its nascent stages. While many existing tools can indicate whether a business is being mentioned and with what frequency, they typically do not provide an assessment of the accuracy of those mentions. At present, verifying the factual correctness of AI-generated information about a specific location largely remains a manual endeavor, requiring direct human review.
As AI technology matures, it is plausible that more sophisticated automated auditing capabilities will emerge. However, for the immediate future, the process of auditing AI-generated content is manual. When inaccuracies are identified, the initial and most impactful action is to optimize the digital assets that businesses directly control. This foundational step is critical for influencing the data AI systems learn from and subsequently disseminate.
Conclusion: Embracing Proactive Accuracy in the Age of AI
The findings from Searchable’s tests serve as a critical wake-up call for businesses operating in the local market. The pervasive inaccuracies in AI-generated information, particularly concerning fundamental details like business location, pose a tangible risk to customer trust and business reputation. The shift towards AI-driven search necessitates a fundamental re-evaluation of how businesses manage their online presence, moving beyond traditional ranking metrics to focus on the accuracy and comprehensiveness of the information AI systems are consuming and disseminating.
By adopting a proactive auditing strategy, meticulously verifying information across multiple AI platforms, and diligently correcting inaccuracies in their own digital assets, businesses can begin to mitigate the risks associated with AI misinformation. Furthermore, a consistent and robust online presence remains the bedrock upon which accurate AI representation is built. As the AI landscape continues to evolve, businesses that prioritize data integrity and proactive monitoring will be best positioned to navigate the complexities of AI search and ensure they are accurately represented to a growing audience of AI-assisted consumers. The future of local business discovery hinges on the ability of both AI developers and businesses themselves to foster an environment of trust through verifiable and accurate information.
