According to figures shared by Searchable Retail Focus, AI tools returned at least one incorrect fact across 64% of UK high street retailers in a vendor-led test. The most common mistake was placing businesses in the wrong zip code.
These tests point to something that many marketing teams don’t track. When AI systems like ChatGPT or Google’s AI Mode answer questions about one of your locations, you can’t find out what they’re saying about your business unless you check it manually. They could confidently claim a business is closed, list services they don’t offer, or provide a false address.
We’ve covered how to gain visibility in AI search, including Dan Taylor’s work tracking AI prompts. This article addresses the other half of the problem. It’s about what AI is already telling people about your locations and how you can find out before a customer does.
From the rankings to the description
Traditional local search offers customers multiple comparison options, such as: B. a map pack, reviews, your website or that of a competitor. Then the customers make their choice. AI search combines this process into a single synthetic answer that fully describes a company before the customer even visits the website or company profile.
Being described is not the same as being classified; Rankings may change and descriptions may not always be accurate. Additionally, this visibility varies depending on the type of query.
Whitespark’s analysis found that AI overviews appear in 15% of direct searches with local intent, such as: [personal injury lawyers in Phoenix]. For informational questions, they appear 92% of the time, and for hybrid questions, such as whether to hire a lawyer after an accident, they appear 97% of the time. In the same study, local packages appeared in 93% of direct searches with local intent. Overall, AI overviews were more common for informational and hybrid questions than for direct searches with local intent.
SOCi’s Local Visibility Index analyzed over 350,000 locations and found that ChatGPT recommended 1.2% of them. According to SOCi’s own measurements, this is a far lower rate than the appearances of the same brands in Google’s local 3-pack. BrightLocal’s Local Consumer Review Survey found that 45% of consumers now use ChatGPT or similar tools for local business recommendations, up from 6% a year ago.
What the answers get wrong
Searchable released two sets of test results to the trade press this month. In tests with 165 London companies, Searchable tested ChatGPT, Gemini and Perplexity with 13,365 questions about services, contact information, size and incorporation dates, then compared the answers to Companies House records and official profiles.
According to CMOTech, 93% of these companies had at least one fundamental fact wrong or missing. Half of smaller companies received at least one false fact, compared to 32% of larger companies. A second test examined UK high street retailers with over 72,000 questions. Retail Focus reported that one in 16 answers was incorrect. Incorrect zip code errors occurred 1 in 10 times, even when the city was specified in the prompts.
Searchable co-founder Chris Donnelly told Retail Focus:
“If a smaller brick-and-mortar retailer’s online visibility is focused primarily on its website and a Google Business listing, that’s a relatively thin trail of information for AI systems to learn from and represent in their responses.”
The error categories match what business owners have described in Google’s support forums. They report that false information is presented to them with confidence, and some say the errors harm their business.
The blind spot
Traditional search leaves a trail of data. Impressions and clicks are tracked in Search Console; Rankings vary depending on the tracker; and every drop often leads to an investigation. However, there is no report that notifies businesses if an AI response gives incorrect hours for your location or claims your business is closed.
Traditional search doesn’t tell you how you’re being presented, but that doesn’t have to be the case. It directs customers to sources they can see and evaluate for themselves, such as your listing, your website, and reviews with attached data. AI summarizes these sources into a single written response, and the customer reads it as fact. If it’s wrong, there’s no ranking position to watch and no traffic drop to track because the error is in text that was never shown to you.
Major consumer AI tools generally do not provide companies with location-level alerts when their information is incorrect. Google’s documentation states that AI responses may contain errors, and individual AI overviews include a feedback link. This feedback is intended to correct errors, but is not a proactive monitoring system. It only works if someone notices the error first.
5 systems, not 1
The testing process can get complicated on different AI search platforms because they don’t work in the same way. AI Overviews now appear as part of regular Google search results, AI Mode is a conversational experience directly within Google Search. Gemini acts as a separate AI assistant from Google. Meanwhile, ChatGPT and Perplexity are separate products offered by OpenAI and Perplexity AI.
Each of these tools processes, selects, and combines information differently, so a single question may receive different answers depending on the platform used. This variation can also be seen in Searchable’s retail data. Perplexity was found to provide inaccurate answers 10% of the time, compared to 5% for Gemini and 4% for ChatGPT.
Testing one system says little about the others. A location checked out in AI overviews may still be described incorrectly in ChatGPT.
How to test what customers see
Start by thinking about questions your customers might have, such as opening hours, services, or whether a location is good. Write these questions on a standard list so that each location is checked in the same way. Complete the questions via AI Overviews and AI Mode in Google as well as ChatGPT, Gemini and Perplexity. If possible, test without saved conversation context or personalization and keep track of the prompts and responses you receive.
Since the answers may be different each time, ask the same questions several times. Organize what you find into categories such as factual errors, missing information, or perception problems. Treat factual errors and omissions separately. Sentiment and recommendation order are reputation issues, not easy fixes. Pay particular attention to correcting errors that could prevent someone from visiting, such as incorrect opening hours or a branch reported as closed.
When citing sources, make sure the information is accurate. Correct the information under your control or request corrections as necessary. Check back regularly because if you fix a source today, the AI answers won’t update automatically. Use a consistent set of questions and protocol to simplify the process, especially if you have many locations. Change the frequency of checking based on the number of locations and how frequently details change.
What to do if the AI makes a mistake?
Finding the error is the first step. Fixing the problem requires carefully adjusting the inputs these systems rely on and then checking to see if the responses improve. Remember: every adjustment brings you closer to better, more reliable results.
If an incorrect time was provided in a reply, confirm that your Google business profile, your website, and your own location pages match each other and the reality. Conflicting details are one of the most obvious reasons an answer is wrong. The same name, address and phone number The consistency of numbers that local SEO has always required is the foundation and becomes more complex when you operate multiple locations.
After you correct the sources you control, look at the sources you don’t own. If a reply mentions your company, open it and check the date and actual content. This could involve retrieving the information from an old directory site, a review listing, or a page about a business with a similar name. Resolving these issues will likely require direct contact.
This is also where the size difference that Donnelly pointed out starts to make sense. Smaller companies made more mistakes in Searchable’s tests, and he found that a thin trail of third-party information leaves these systems with fewer opportunities to work. Our own reporting on what’s related to ChatGPT quotes comes to the same conclusion. A more comprehensive and consistent presence in the places these systems read gives them better material with which to describe you.
If you think accuracy isn’t the issue and you’re more concerned with getting recommended in the first place, then this is the visibility side. Dan Taylor wrote for Search Engine Journal about tracking how AI represents your brand over time and how discovery changes as search becomes personalized for each user. These are the starting points for the question “How do we win here?”
Looking ahead
AI response monitoring tools are still in their infancy. Most can tell you if you are mentioned and how often, which is not the same as telling you whether the mention is accurate.
To find out what an answer actually says about a place, you still have to read it yourself. As these tools evolve this may change, but currently the review is manual and if you find an error, the first step is to optimize the sources you control.
Additional resources:
Featured Image: Natalya Kosarevich/Shutterstock
Follow us on Facebook | Twitter | YouTube
WPAP (907)
