I've been running this test on prospects' businesses for a while now. The answers are almost never clean.
Last week's post was about schema cannibalization, your own site confusing Google about itself. This is the bigger version, playing out across the entire AI ecosystem instead of on one site. It's called entity confusion. And once you know to look for it, you can't unsee it.
What entity confusion actually is
Here's what's happening underneath. When AI systems can't tell two similar businesses apart (same name, same region, overlapping services), they blur them together. Reviews get misattributed. Pricing gets cross-cited. The customer-facing answer gets built from the wrong source material, and the buyer doesn't know.
Google has had over a decade to build its Knowledge Graph, its big database of "things in the world" and how they connect. When you search for a business on Google, it generally knows which one you mean, because it's been doing this work for that long.
AI systems are starting almost from scratch. They synthesize answers from web content, training data, and live retrieval, all at once. When two businesses share a name, a region, or even a similar service description, the systems often blur them together. Your reviews end up cited under a competitor's name. A competitor's pricing ends up cited under yours. A 1-star review someone left for a different business in 2017 becomes "your" reputation in a Perplexity answer in 2026.
The cleanest way to think about it: schema cannibalization is your site confusing Google about itself. Entity confusion is the whole open web (and the AI systems reading it) confusing your business with someone else's.
Why this is a business problem, not a technical one
The behavior change that makes this expensive isn't on the AI side. It's on the buyer side.
People aren't clicking blue links the way they used to. They're asking ChatGPT, Claude, Perplexity, and Gemini for recommendations and trusting the answer that comes back. When that answer contains entity confusion, your business loses the opportunity before the buyer ever visits your site. You don't get the analytics signal. You don't get the form fill. You get nothing, and you don't know why.
Here's what I've watched happen:
- A buyer asks an AI for a recommendation in your category. The AI confuses you with a lower-rated competitor and points the buyer elsewhere.
- A buyer asks about your pricing. The AI cites a competitor's pricing as if it were yours, and they show up either over- or under-prepared when they finally contact you.
- A buyer asks for reviews. The AI surfaces a 1-star review from a different business with a similar name.
- A journalist or B2B procurement team researches you. The AI hands them a Frankenstein profile stitched together from you and two other businesses.
You cannot opt out of this. AI systems will keep answering questions about your business whether you participate in the conversation or not.
The 5-minute self-audit
Run all five. Jot down what you find. Five minutes, tops.
| Test | What to do | What you're looking for |
|---|---|---|
| The ChatGPT test | Ask ChatGPT five basic questions about your business: founding year, location, services, leadership, key reviews | Any factual error, anything generic, anything that doesn't match your reality |
| The Perplexity test | Same five questions in Perplexity. Click through every source it cites | Whether the sources are actually yours, or whether some belong to a similarly named business |
| The AI Overview test | Search your business name on Google. If an AI Overview appears, read it for the same kinds of errors | Wrong facts, sources that aren't yours, conflated information |
| The competitor mashup test | Ask: "Compare [your business] and [competitor with similar name]." | Whether the AI can cleanly distinguish you. If it blurs the two, you have an entity problem |
| The review attribution test | Ask: "What do customers say about [your business]?" | Whether the reviews it summarizes are actually yours |
If you get clean answers across all five, you're in better shape than most businesses I audit. Find issues in two or more and this is now on your priority list.
The hard part isn't finding entity confusion. It's that nobody on most marketing teams is checking for it.
What actually fixes it
This is where most posts about AI search wave their hands and call it a day. The real fixes split between work on your own site and reinforcement signals across the open web. Both matter. Neither is optional.
Start with your knowledge graph signals. Your Organization schema needs to say, in plain code, exactly who you are, where you are, when you were founded, and how you connect to your other public profiles. The sameAs property is the most underused piece of this. It's where you tell AI systems that your LinkedIn page, your Crunchbase profile, your Wikipedia entry, and your domain are all the same entity. Without it, the systems are guessing. And they don't always guess in your favor.
Get your NAP consistent. Name, Address, Phone Number. The exact same way across every directory, every social profile, every listing you've ever submitted to. "Smith & Co." on one, "Smith and Company" on another, "Smith Co" on a third. That's exactly the gap AI systems fill with whatever they want. Unglamorous work. Compounds heavily.
Get mentioned on authoritative external sites. AI systems trust mentions on independent, well-known sources more than mentions on your own domain. Industry publications. Trade resources. Podcast interviews. Guest articles. And LinkedIn especially, where AI systems already pull more of their B2B answers from than most marketers realize. For most B2B brands, off-site mentions move the AI-visibility needle harder than anything published on your own site.
Disambiguate hard in your own copy. If your brand name has a common-noun meaning, or a near-twin competitor, your About page and homepage need to make the distinction obvious in plain English. Founding year, headquarters, industry, leadership. All of it should be on the page where buyers and AI systems land. Don't make either of them work to figure out who you are.
Re-check on a regular cadence. Entity confusion isn't a one-time fix. AI systems retrain. The web changes. New competitors launch every week. Whatever you fix this quarter needs to be re-checked next quarter. Bake the 5-minute self-audit into a recurring task and you'll catch new confusion before it costs you a deal.
Why this matters more right now
LinkedIn is now the #2 most-cited domain in AI search, behind only Reddit. That means a single inaccurate LinkedIn description of your company has more weight in an AI answer than three solid blog posts on your own domain. The platforms AI systems trust most aren't always the ones you've been optimizing for.
What that means for content strategy: if all your AI visibility work is happening on your own domain, you're pulling one lever in a system that has at least four. The other three (where your competitors are quietly racking up entity-reinforcement signal) are mostly off your site entirely.
Most marketing problems that quietly cost businesses real money are invisible from the dashboard. You can't see entity confusion in Google Analytics. It doesn't trigger errors. It just gradually erodes how AI systems describe your business to potential customers, and the buyers who would have found you go elsewhere instead.
The marketers who'll matter over the next five years are the ones who can see these invisible problems, have the technical depth to fix them, and treat AI search as a real visibility channel instead of a buzzword on a discovery call.
Next week I'll close out this short series with a post on the discipline most agencies skip, the one reason your monthly report can't tell you what's actually working.
Frequently asked questions
Entity confusion is when AI systems like ChatGPT, Perplexity, and Google AI Overviews fail to distinguish between similar businesses or concepts. The result is misattributed reviews, wrong pricing, conflated services, and recommendations that send buyers to a competitor. It's different from schema cannibalization, which is your own site confusing Google about itself. Entity confusion is the open web (and the AI systems reading it) confusing your business with someone else's.
There isn't a single setting that controls this, but the fundamentals are knowable. Clean, complete schema on your site. Consistent name, address, and phone across every directory and profile. Authoritative external mentions on sites AI systems already trust, especially LinkedIn, Wikipedia, and industry publications. Aggressive disambiguation in your own content so AI systems aren't guessing about basic facts. The brands AI systems recommend are the ones whose identity is unambiguous everywhere they appear.
Open ChatGPT and ask five questions: when was your business founded, where is it headquartered, what services does it offer, who runs it, and what do customers say about it. Repeat the same set in Perplexity and click through every source it cites. Repeat the brand-name search in Google and read any AI Overview that appears. If any of those answers contain factual errors or sources that aren't actually yours, you have entity confusion to fix.
sameAs is a property in schema markup that lets you list other URLs representing the same entity. For a business, that means your LinkedIn Company Page, your Crunchbase profile, your Wikipedia or Wikidata entry, your GitHub, and any other authoritative source that represents the same business. Adding sameAs to your Organization schema tells AI systems and search engines that all of these signals refer to one entity, not several.
Recent research has LinkedIn as the second-most-cited domain in AI search, behind only Reddit. AI systems weight LinkedIn content heavily when answering questions about businesses and people. For most B2B brands, a clean, accurate LinkedIn Company Page and a consistent personal profile for the founder or CEO carry more AI-search weight than a blog post on the company domain.
No, and that's an important expectation to set. AI systems are constantly retraining and reindexing. New competitors launch, old data persists, and the open web keeps changing. Entity reinforcement is ongoing work, not a one-time project. The teams who get this right are the ones who treat it as a recurring quarterly check, the same way they treat any other visibility audit.
Yes, in most cases. The same signals that disambiguate your business for AI systems (consistent NAP, strong Organization schema, sameAs declarations, authoritative external mentions) also reinforce your entity in Google's Knowledge Graph. The work compounds across both traditional search and AI search, which is why it's one of the better uses of marketing budget right now.