Your AI Is Lying About Your Business Hours

A customer asks ChatGPT when you close on Sunday. ChatGPT tells them 5 PM with complete confidence. You close at 2 PM. They show up at 3:30, find a locked door, and book your competitor before they’ve pulled out of your parking lot. You find out about this approximately never.

An AI business information accuracy audit is a systematic check of what AI tools (ChatGPT, Gemini, Claude, Google Search) are actually saying about your hours, address, phone number, and services. It takes about 30 minutes, costs nothing, and most businesses find at least one factual error the first time they do it.

01Why AI Gets Your Hours Wrong (And It’s Not Always Your Fault)

AI tools don’t call your front desk. They pull business data from wherever they can find it: Google Maps, Yelp, third-party directories, old review sites, forum posts, aggregated data scrapers. Some of those sources were accurate in 2021. Some were never accurate. Once that data gets baked into a model’s training set, it tends to stay there long after you’ve updated your actual hours, moved locations, or changed your phone number.

According to Search Engine Journal, AI tools serve incorrect location data, hours, or service details in nearly half of all local business answers. That’s not a rounding error. That’s a coin flip on whether the information a customer gets about your business is accurate.

The small business version of this is worse than average. Big chains have PR teams, brand monitoring software, and entire departments whose job is keeping their data clean. You have whatever you last updated on Google.

Not all AI tools work the same way, and the distinction matters. Gemini reads Google Maps directly, which is why SOCi’s 2026 research found it hitting 100% accuracy on business profiles. ChatGPT and Perplexity average around 68%, because they’re drawing on training data with a cutoff and no live feed to your actual listing. So when your hours changed in January and you updated Google Business Profile immediately, Gemini knows. ChatGPT is still running on whatever it learned before September 2024.

And this matters more than it probably should: AI Overviews now appear in 68% of local searches, according to Whitespark’s Q2 2025 data, compared to 39% for traditional local packs. AI isn’t a secondary discovery channel anymore. For a lot of searches, it’s the primary one.

02Where to Actually Check What AI Is Saying About You

This is the part most guides skip, so let’s be specific. Open each of these and run the same queries. You’re looking for your business name, address, phone number, hours, and services listed anywhere in the response.

The queries to run

• “[Your business name] hours”
• “[Your business name] phone number”
• “[Your business name] address”
• “[Your service type] near [your city]” … this checks if you show up at all, and what it says when you do
• “Is [your business name] open on [day]?”

Where to run them

• ChatGPT (GPT-4): Free tier works. This is your most important check, given how many people use it.
• Google Search: Look at the AI Overview at the top, and separately at the Knowledge Panel on the right. Those can contain different data.
• Google Maps: Search your business directly. This is the source Gemini reads, so whatever’s here is what Gemini will say.
• Perplexity: Pulls from live web data with citations. You’ll be able to see exactly which source it’s drawing from, which is useful for diagnosing where bad data lives.
• Claude: Lower risk for local business data (it tends to hedge more and cite less), but worth a quick check.

Document everything. Screenshot each response. If you find conflicting information between tools, that tells you your source data is inconsistent across directories. Which brings us to what actually fixes it.

03The One Source of Truth Google (and Everyone Else) Actually Trusts

Your Google Business Profile is not just a listing. It’s the data layer that feeds Gemini directly, influences what Google’s AI Overviews say about you, and sets the baseline that every other directory eventually tries to match. If it’s wrong or incomplete, everything downstream is wrong. If it’s right, you’ve solved the biggest part of the problem.

Only 27% of phone numbers provided by AI tools match a business’s actual Google Business Profile, according to Seer Interactive’s 2025 research. The other 73% have a mismatch somewhere: either the profile is wrong, the AI pulled from somewhere else, or the directories feeding the AI have stale data. Those three causes have the same fix.

The basics that need to be exactly right:

• Business name: Exactly as it appears on your signage and website. No keyword stuffing, no location names appended unless they’re legally part of your business name.
• Address: Down to the suite number. A single wrong digit has caused three different AI tools to show three different addresses to potential customers. Not a hypothetical.
• Phone number: Your primary number, ideally the one that’s also on your website. Consistency matters more than which number you pick.
• Hours: Including special hours for holidays and seasonal changes. Update these proactively, not after a customer complains.
• Primary category: The most specific category that describes what you actually do. This affects which “near me” searches surface your profile.
• Services: List them explicitly. AI systems use this to answer “does [business] do X?” questions.

Google Business Profile is free and takes about an hour to audit properly. Do that before anything else. Everything else is downstream of this.

04Schema Markup: The Unsexy Thing That Actually Works

Schema markup gives your website a machine-readable layer that states your business information in a format crawlers and AI systems can parse without ambiguity. Your website probably already describes your hours and address in text somewhere. Schema markup is the version of that information a machine can actually read reliably, instead of guessing from context. Think of it as the difference between leaving a note on the fridge and filing it in a labeled folder that a very literal-minded assistant can find in the dark.

For local businesses, the relevant schema type is LocalBusiness (or a more specific subtype: Restaurant, Plumber, MedicalBusiness, etc.). The fields that matter most for AI business information accuracy:

• name … your exact business name
• address … structured as streetAddress, addressLocality, addressRegion, postalCode
• telephone
• openingHoursSpecification … this is what tells AI systems your actual hours, day by day
• url
• geo … latitude/longitude, which helps with map-based AI tools

You don’t need a developer to add this. Google’s Structured Data Markup Helper walks you through it visually, and most CMS platforms (WordPress, Squarespace, Wix) have plugins or built-in fields that generate schema automatically. The output is a block of JSON-LD that goes in your site’s <head>. Paste it into Google’s Rich Results Test to confirm it’s reading correctly.

The reason schema markup matters specifically for AI accuracy: it gives AI systems a high-confidence source directly from your own domain. When your website, your Google Business Profile, and your structured data all agree, there’s no reason for an AI tool to trust a 2019 Yelp listing over your own authoritative data. Consistency is the argument. Schema is part of making it.

05How to Stop the Bleeding While You Fix It

If you found errors during your audit, fix them in order of reach, not order of annoyance.

1. Fix your Google Business Profile first. This propagates to Gemini and Google’s AI Overviews immediately. Biggest reach, fastest update.
2. Update your website’s schema markup second. This is your owned data source: no approval process, no wait time.
3. Audit the major directories third: Yelp, Apple Maps, Bing Places, Facebook, and any industry-specific directories relevant to your business. Inconsistent data across these is what makes AI tools uncertain about which version to trust.

For monitoring going forward, a few tools that actually help:

• Google Business Profile’s built-in suggested edits alerts: Turn these on. When a user or algorithm suggests a change to your profile, you want to know before it goes live.
• Yext or Moz Local: These sync your business information across dozens of directories from one dashboard. Not free, but if your address is listed twelve different ways across twelve different platforms, the time savings are real.
• A recurring calendar reminder to query ChatGPT and Perplexity quarterly: Takes ten minutes. Do it after every significant change to your hours, address, or services.

If you find AI spreading actively wrong information and need to flag it: ChatGPT has a thumbs-down feedback mechanism on responses, Perplexity has source correction options, and Google’s Knowledge Panel has a “suggest an edit” link. None of these are fast. Plan for weeks, not days, which is exactly why fixing the source data is the only real answer.

The stat that puts the stakes in proportion: 45% of consumers used AI tools for local business recommendations in the past year, up from 6% the year before, according to BrightLocal’s 2026 data. That’s not a trend you’re watching develop. That’s already your customer base, getting information about you from a source you’ve never touched.

06This Is Already Happening

AI systems are answering questions about your business right now. Hours, address, phone number, services you may or may not actually offer. You didn’t authorize any of it, and you don’t get a notification when it happens. You find out when a customer mentions it, or you don’t find out at all because they just didn’t show up.

The audit takes 30 minutes. The Google Business Profile fix takes an hour. Schema markup is an afternoon. Run the audit, fix what’s wrong, and set a quarterly reminder so it stays fixed.