Is Your Brand in the Data That Trains ChatGPT?

Why your Common Crawl footprint matters

There's one sentence from the people who run Common Crawl that I think every business owner should have taped to their monitor. Here it is: "If you are not in the crawl, you are not in the model." Let me translate that, because it's bigger than it sounds. The AI models everyone's suddenly asking for recommendations — ChatGPT, Gemini, Perplexity — they weren't trained on some mysterious, unknowable soup. A massive share of what they learned came from one specific, public, nameable source. And there's a real, checkable answer to the question "is my business in it." Almost nobody has looked. Today we're going to look — because being in that data is quietly becoming the new page one, and being left out of it is a kind of invisible that's very hard to come back from.

Training data and the new page one

I'm Brian Winum, this is MAXVolume, and I want to start by making "in the model" mean something concrete, because it gets thrown around like magic and it isn't.

When a model is built, it's trained on an enormous slice of the web — frozen at a point in time. Whatever it absorbed about your industry, your competitors, and you becomes its baseline knowledge. So when a buyer later asks it cold, "who are the good firms for this," the names it reaches for are the ones that were well-represented in that training data. Not the ones with the slickest site today. The ones that were in the corpus when the thing was built.

And that's why I keep calling this the new page one. For twenty-five years, ranking on page one of Google was really just a proxy for legitimacy — it meant the market recognized you as a real player. That signal is moving. The new proxy for "you're a real, established player" is whether you're baked into the data these models learned from. If you're in there, you're in the room when the AI makes its shortlist. If you're not, there's no page two to settle for — there's just the set of names the model learned, and the set it never did.

Common Crawl and the research behind it

Now here's the problem everyone hits: you can't call OpenAI and ask to see their training set. It's a black box. So people assume this whole layer is unknowable and they give up on it. It's not unknowable. Because the single biggest ingredient in that training data has a name, and it's public.

It's called Common Crawl — a nonprofit that's been archiving the open web since 2008. We're talking about a dataset north of nine petabytes. And it is, by a wide margin, the foundational training corpus of the AI era.

A Mozilla study last year found at least sixty-four percent of major language models were trained on filtered Common Crawl data. For GPT-3, more than eighty percent of the training tokens traced back to it. So you can't read the model's mind — but you can check the thing the model was built from. "Are you in the AI" turns into a concrete, answerable question: are you in Common Crawl? And that is a question you can actually get an answer to this afternoon.

Has the crawl captured your site?

When you go look, there are really two things that matter, and this is where it gets useful. The first is simple inclusion. Has Common Crawl actually captured your pages at all? For a surprising number of businesses — especially anyone who ever pasted in one of those "block the AI bots" snippets, or who's got a thin or badly structured site — the answer is no, or barely. And if the crawl never captured you, you were never a candidate to be learned in the first place. That's the door being shut before anything else can happen.

Web graph authority and trends

The second is authority within the crawl — and this is the part most people don't even know exists. It's not enough to be a stray page in a nine-petabyte ocean. Common Crawl also carries a web graph — essentially a map of how connected and how central your site is relative to everything else. Your Harmonic Centrality, your PageRank inside that graph. Being present is table stakes; being prominent is what makes you likely to actually be learned and retained instead of lost in the noise. Presence gets you in the building. Centrality gets you remembered.

And then there's a third dimension across time — the trend. Is your footprint in each new crawl release growing, flat, or quietly fading? Because this corpus refreshes, and so do the models trained on it.

The free AI Training Data Visibility Check

So this is exactly why we built a free tool called the AI Training Data Visibility Check. You put in your domain — no sign-up, nothing to buy — and it goes straight to Common Crawl and pulls those three signals back for you: are you in the index, what's your authority inside the web graph, and which way is your trend moving across releases. Three signals, one blunt answer: can the AI actually see your brand in the data it learns from.

What the checker can—and cannot—tell you

Now let me be honest about what it is and what it isn't, because you've heard me do this every episode and I'm not going to stop now. This is a proxy. It is not a magic window into OpenAI's weights — nothing on earth is, and anybody who tells you their tool reads what ChatGPT was trained on is selling you something. What it does is measure the single largest, best-documented input to these models. That's the most honest signal available, and it's a lot better than guessing.

And it is strictly a diagnosis — the checker tells you the score, it doesn't change it. It fixes nothing. It just ends the guessing.

One more honest note, same as last episode: you can't edit a model that already exists. What you influence is what the next crawl captures and the next model learns. So the footprint you build now is a long game — planted today, harvested at the next generation. Which is exactly why checking where you stand now matters so much.

The free check versus the paid AI Readiness Audit

The free check answers "am I in the data." The bigger question — "okay, I'm thin or I'm invisible, now what, and in what order" — that's a different, deeper piece of work. That's our paid AI Audit, the full scored assessment we talked about a few episodes back: the eight categories, the prioritized roadmap, and the AI Readiness Score that comes out of it. The free Visibility Check is the front door. The Audit is the blueprint.

I want to keep those straight for you, because they're two different tools doing two different jobs — one tells you if there's a problem, the other tells you exactly what to do about it.

Where to start and next steps

So here's where I'll leave it. Page one was about being found. Your training-data footprint is about being in the corpus the machine was built from — and that's the deeper, longer game, which means every day you don't know where you stand, the gap can quietly widen without a single number on your dashboard moving. So go get the number.

The free AI Training Data Visibility Check lives at a-i-o dot maxplaces dot com — domain in, three signals back, no sign-up. And if it tells you what it tells a lot of people — that you're thinner in there than you'd like — the full audit and the roadmap to fix it are at audit dot maxplaces dot com. Start with the free one.

Find out if you're in the crawl, because if you're not in the crawl, you're not in the model. I'm Brian Winum. Thanks for turning it up. I'll see you next time.