Free Brand Fact Consistency Check: Find Contradicting Facts on Your Own Site

The check answers one question: does your own site say the same thing about you everywhere? If your pricing page says $49, your footer says $39 and an old launch post says $29, a crawler that fetches all three gets three answers.

That matters because you control it. Most of what happens inside an AI answer is out of your hands, but your own pages are not. If you want the background on how site errors show up in answers, read whether mistakes on your own website can cause AI errors.

The output is a short table: each fact, its correct value, and every URL where it appears with a different value. That table is your fix list.

Start with a fact sheet, not a crawl. Write each fact once, with the value you want the world to repeat. Keep it to facts a buyer or an AI answer would actually state.

• Pricing: every plan name, price, billing period and currency.
• Founding year: one year, and whether you mean incorporation or launch.
• Location: headquarters city, plus any office list you publish.
• Product names: current names, and the old names you have retired.
• Ownership and size: parent company, acquisitions, headcount if you publish it.
• Contact basics: support email, phone number, legal entity name.

Add a column for retired values. Old prices and old product names are what you will be searching for. The guide on consistent brand facts covers how to keep this sheet current after the first pass.

Say you run a hypothetical project tool at example.com with a Starter plan at $49. Here is the process.

1. Search your own site for each value. In Google, run site:example.com "$39" and site:example.com "founded in". Repeat for each retired price, old product name and old city.
2. Fetch the raw HTML of key pages. AI crawlers often read only the HTML the server sends. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files but do not execute them. Check what they see:

curl -s https://example.com/pricing | grep -i -E "\$[0-9]+"

3. Compare against the rendered page. Open the same URL in a browser. If a price shows in the browser but not in the curl output, it only exists after JavaScript runs.
4. Check the footer and header templates. One stale line in a footer repeats on every page. Search the HTML of any page for your founding year and address.
5. Check markup blocks. Structured data can state a different value from the visible text:

curl -s https://example.com/ | grep -i -E "foundingDate|price|addressLocality"

6. Check old content. Press releases, launch posts, changelogs and PDFs often carry the old price or name. Note each URL.
7. Record every mismatch in your fact sheet: URL, location on the page, the wrong value.

Sort mismatches by how much damage a wrong answer would do.

• Fix first: a wrong price, a wrong plan name, or a fact that only appears after JavaScript. A wrong price costs you deals. Content that needs JavaScript may never reach most AI crawlers.
• Fix next: footer and template errors, because they repeat site-wide.
• Fix or annotate: old posts. Update the number, or add a dated line saying the price changed.

When a structured data block disagrees with the page, make it match the visible text. Do not add markup expecting citations. Google's AI optimization guide says you don't need special markup to appear in its generative AI features. The point is removing a contradiction, not adding a signal.

Then state each fact once, plainly, in the server-sent HTML. A line like "Starter costs $49 a month" gives an engine a sentence it can quote. A 252,000-trial study (arXiv:2605.25517) found that including explicit price information helped consistently, but it ran on content already inside the model's context. That is no promise that engines will pick your page. For pricing specifically, see whether to publish your prices for AI answers. Dates have their own check: use the date consistency check for lastmod, headers and visible dates.

A clean site is necessary groundwork, not a result. Be clear about the gaps.

• It does not see third-party pages. Directories, review sites and old articles can carry wrong facts. Engines that cite those sites may repeat them. Read what AI does when sources disagree, then work through asking third-party sites to fix wrong facts.
• It does not touch training data. When a model answers without searching, it relies on what it learned before its knowledge cutoff. Nothing you change this month reaches that answer.
• It does not confirm access. A page with perfect facts still fails if robots.txt or a firewall blocks the crawler.
• It does not measure answers. To see what engines actually say, ask them. The free checker on this site shows whether engines name you for up to three buyer questions, as a one-time snapshot.

Engine behavior described here is as of October 2026 and changes often.

The manual process works, but it is easy to skip a template or a markup block. The AI Content Checklist gives you a structured list to work through, so each page gets the same review. Keep your fact sheet beside it as the single source of truth.

Once your own site agrees with itself, check whether AI answers get your numbers right with the free AI pricing accuracy check. Rerun the consistency check after every pricing change, rebrand or office move.