Yes, AI-written town pages can contain invented local details. A landmark that sits in the next county, a neighborhood the page says you serve but you never visit, a drive time that is off by half an hour. Often a customer is the one who points it out.
The fix is a short check, done on one page first. Every named place, distance, service promise, photo and contact detail gets confirmed against a reliable map or your own records. Anything you cannot support comes out.
You can run that check alone this week, and ten minutes on a single page will usually show you how bad the rest are. After that comes the bigger decision for each page: fix it, merge it, or remove it.
Key Takeaways
List every place name, distance, service claim, photo and contact detail on a single town page, confirm each against a map or your own records, and remove whatever you cannot support.
It does not prove you serve the place, that a claim is current, or that a photo shows the town it is captioned with.
Fix it when you really serve the town and can add accurate specifics; merge or remove it when it is a city-name swap, a duplicate, or an area you do not serve, and redirect any merged URL.
How do I check one town page for invented details in ten minutes?
Pick the town page you are most worried about and open a blank document next to it. You are building a correction list, so do not edit the page yet. Work down the page and copy every claim into the list, then confirm each one in this order.
- Places: Copy every named landmark, road, neighborhood, park and school. Look each one up on a reliable map or the town's own website, and confirm it sits inside the town the page names, not just near it.
- Distances and drive times: Recalculate each one from the same starting point the page uses. If the page says "15 minutes from our shop," measure from the shop.
- Service promises: Compare each promise with the area you actually cover and the schedule you can keep. If the page says same-day service and you do not offer it there, that line goes.
- Business details: Check the address, phone number, hours and any license or certification claim against your own records.
- Photos: Trace where each image came from. A stock photo with the town's name in the file name is still a stock photo.
- Customer stories: Confirm each testimonial or project example against a real job or a real review.
- Local rules: Anything about permits, codes or town requirements needs a link to the town or the agency that sets the rule, or it comes out.
Mark each line confirmed, needs clearer wording, unsupported, false or out of date. Google's guidance for business information is plain: reflect your business accurately, with an address or service area that is accurate and precise.
| Claim on the page | Confirm it against | If you cannot confirm it |
|---|---|---|
| Landmark, road or neighborhood | A map and the town's own website | Remove it |
| Distance or drive time | The same starting point on a map | Recalculate or remove it |
| Service promise | Your real service area and schedule | Rewrite to what you do |
| Address, phone, hours | Your Business Profile and your own records | Correct it today |
| License or certification | The issuing agency's record | Remove it |
| Photo | Your own files or a license you hold | Replace it |
| Testimonial or project | A real job or a real review | Remove it |
Two limits are worth keeping in mind. A map confirms geography, but it cannot tell you whether you serve a place or whether a claim is still current. And a photo can show a real street without showing your work. Only your own records settle those.
Why do AI tools make up landmarks, neighborhoods and distances?
AI writing tools build sentences from patterns. OpenAI says language models generate answers from patterns in training data and can produce plausible but false statements, including fabricated citations and incorrect dates, delivered with confidence.
Small local facts are the weak spot. OpenAI's 2025 explanation of hallucinations says next-token prediction is good at fluent patterns but weaker with arbitrary, low-frequency facts, such as a particular person's birthday. A street corner, a neighborhood boundary or the name of a small park is exactly that kind of fact.
The U.S. National Institute of Standards and Technology uses the word "confabulation" for this in its Generative AI Profile, published July 26, 2024. It says generative models approximate the statistical distribution of their training data, which can produce factually inaccurate or internally inconsistent output.
So, a page that reads smoothly tells you nothing about whether it is right. A cheap content service that never opened a map can produce the same errors, which is why the check works the same either way.
Should I fix, merge or remove each town page?
Once the correction list is done, decide each page on what it can honestly say. The deciding question is whether the page can answer something a visitor in that town needs, in a way your other pages do not.
Fix the page when you really serve the town and can add accurate, specific information, such as how you cover the area, relevant work you have done, real customer evidence or local constraints that affect the job.
Merge or remove it when the page is a city-name swap with largely identical copy, duplicates another page, targets an area you do not serve, makes unsupported local claims you cannot replace with real ones, or cannot say anything that a broader regional page does not.
| What the page looks like | Decision |
|---|---|
| You serve the town and can add real specifics | Fix it |
| Same copy as other pages with the town name swapped | Merge into one stronger page |
| Duplicates another page's answer | Merge into the stronger page |
| Covers an area you do not serve | Remove it |
| Local claims you cannot support and cannot replace | Remove it, or rebuild it from facts |
When pages are merged, send the old address to the most relevant surviving page with a redirect, and update your internal links and sitemap. If a page should disappear entirely, remove it or block it from search. Google notes that the Removals tool in Search Console is temporary and does not replace removing or blocking the content itself.
Google's core update guidance says deleting unhelpful content can help good content perform better, especially where whole sections were created for search engines rather than people. You may also find you never needed a page for every nearby town. Our posts on service area pages, showing up in nearby towns and the swap-the-name test cover how many pages a business can honestly support.
Does Google treat AI-written town pages as spam?
Google's spam policies say the rules apply whether content is made by AI, by people or by a mix of both. What they target is purpose and value. The spam policies list scaled content abuse, meaning many pages generated primarily to manipulate rankings rather than help people, and that explicitly includes using generative AI tools to produce many pages without adding value for users.
Doorway abuse is a separate item. Google defines doorways as pages created to rank for specific, similar search queries that lead visitors to an intermediate page less useful than the final destination. Its examples include multiple city or regional pages that funnel people to one page, and near-identical pages with no clear browsing hierarchy. Our post on what a doorway page is goes through where the line sits for a local business.
Put simply, a single accurate page about a town you serve is not the target. Hundreds of near-copies made to catch every search variation are. Google's helpful content guidance also asks for factually accurate content, and treats automated generation without human oversight or curation as little or no effort when quality is assessed.
What can I stop worrying about?
Some things feel urgent and carry little weight. You can set these aside while you work on the claims themselves.
- Word count: A longer page full of unverified claims is a longer page of problems.
- The town name in the URL or headings: It does not show that you serve the place.
- A LocalBusiness schema block: Google's documentation asks for structured data that matches what the page visibly says, so markup cannot make a false claim true.
- A landmark paragraph on every page: A plain, verified explanation of how and when you serve the town is more useful than a list of attractions.
- Whether AI touched the draft: The question is whether each claim holds up, not who typed the first version.
The higher-priority problems are false claims, wrong business information, near-identical pages, and service promises you cannot keep. Fix those first.
When does paid help make sense?
For one page, the ten-minute check above is enough, and you can run it yourself. Paid help starts to make sense when the problem is bigger than one page: dozens of URLs, unclear ownership of the content, redirect work, a content management system that limits what you can change, several service areas, or claims tied to a license that must be exact.
A useful engagement has a clear shape. It includes a claim-by-claim inventory of each page, a fix, merge or remove recommendation for each URL, redirects set up correctly, and a review step so the same errors do not return with the next batch of pages. It does not promise a ranking, because nobody can.
Our content marketing and local SEO services cover that work. A smaller job, such as a handful of pages, is one you can handle yourself with the list above.
Can you trust your town pages?
Pick an answer to begin.
1. What does a map prove about a town page?
2. Two town pages have identical copy except for the town name. What is usually the right move?
3. What do you do with a claim you cannot support?
Frequently Asked Questions About AI-Written Town Pages
Can AI-generated town pages contain made-up facts?
Yes. AI tools can produce confident but incorrect local details, so check every place name, distance, service claim, photo and contact detail before a page stays live.
Is AI-generated content bad for SEO?
Google says its policies apply whether content is made by AI, people or both. The risk comes from many pages made to rank without adding value, or from pages with inaccurate claims, not from AI use alone.
Are location pages important for SEO?
A page for a town you really serve, with accurate and specific information, can answer a real customer question. Pages that only swap the town name rarely do, and you may not need one for every nearby town.
What should I check first on a town page?
Start with named places, distances, service promises, and your address, phone and hours. Then check photos, testimonials and any license claims.
Should I delete every AI-written city page?
No. Keep and correct pages that serve a real customer need and can carry accurate, distinct information. Merge or remove the rest, and redirect merged addresses.
Can I claim a town if I do not have an office there?
A service-area business can describe the area it truly serves without inventing a local office or address. Google also generally expects one Business Profile for the central location with a designated service area.
What This Means for You
Start with one page. Build the correction list from the places, distances, service promises, contact details and photos, confirm each against a map or your own records, and take out what you cannot support. Then give every page a decision: fix, merge or remove. Remember that a map proves geography and nothing more.
Done once, the check leaves you with fewer pages that you can stand behind, claims you can trace to a source, and a customer who finds the landmark where the page said it would be. Adding a verification step before any new page goes live keeps the same errors from coming back.
If you have many URLs or tangled redirects and want help, Web Leveling can work through them with you. Our content marketing team rebuilds pages from facts you can support, and our local SEO work settles which towns deserve a page at all. We work with small and medium businesses across the country and overseas. Send us your list of town pages and we will tell you which to keep, merge or remove.
Terms
Words used in this post
Tap a term to see what it means.
Confabulation. NIST's word for an AI system producing confident but inaccurate or inconsistent output.
Scaled content abuse. Google's term for many pages generated primarily to manipulate rankings rather than help people.
Doorway page. A page made to rank for a specific, similar search that sends visitors on to a less useful page.
Service-area business. A business that goes to its customers instead of serving them at a storefront.
Redirect. An instruction that sends a visitor and search engines from an old address to a new one.
Correction list. A page-level record of each claim, whether it was confirmed, and what to change.
