Here's exactly how we count. All of it.

Ask an AI the same thing 100 times and the names change.

You won't get the same answer 100 times. The companies it names shift from one ask to the next. That's not a glitch. It's how these systems work. They write a fresh answer every time instead of reading off a fixed list.

Why this matters for you: any tool or agency telling you "your business ranks #3 in ChatGPT" is selling you a number that doesn't exist. There is no rank. ChatGPT holds a position #3 for roofers in your city about as much as a bar conversation does.

What we measure is frequency of inclusion. Out of many asks, how often does your company come up? "You showed up in 7 of 10 asks this week, up from 4 last week" is something you can check. "You rank #3" isn't. Every number we report is a frequency with the sample size attached. Never a fabricated rank.

When AI names a local roofer, it leans hardest on your Google profile.

Your hours, your categories, your services, your reviews, your location. For Google's own AI Overviews, that profile is basically where the answer comes from. Let it go stale and you're close to invisible, no matter what else you've done. It's also the most fixable thing on the list. So we start there.

Under about 4 stars, AI leaves you out — no matter what else is right.

This is the finding that surprises most roofers. It's also the one that decides whether we can help you at all. On traditional Maps results, a 3.7-star outfit down the street still shows up. AI doesn't work that way. It appears to apply a rating floor. Above it, more stars help. The floor itself is pass/fail.

What this means for you: if your rating is below roughly 4.0, the honest answer is that no amount of optimization gets you reliably named until that number comes up. We'll tell you that to your face. The alternative is taking your money and fiddling around the edges, and we won't. Some weeks the first real job is your reviews, not your AI visibility. If that's you, we'll say so.

What others say about you counts more than what you say.

AI leans hard on what shows up about you elsewhere — Reddit, YouTube, industry directories, review sites. Your own website counts. But the consistent picture of you across the web counts more. We call this outside proof: real mentions of you on the sites AI reads. For roofing, video pulls unusual weight, and the bar to clear is lower than you'd think. This is the slowest, least flashy work we do. It's also the part that lasts. That's where durable AI visibility actually comes from.

Today AI search is small. It's growing fast, and the slots are few.

The honest size of it: across the web today, AI tools drive a tiny fraction of the visits traditional search does — the gap is on the order of a hundredfold. Anyone telling you AI search is already a flood of customers is overselling.

But the intent is high and the slots are scarce. A homeowner asking an AI "who's the most reliable roofer in town" is about as deep in buying intent as it gets. The answer names a handful of companies. For something as big-ticket as a roof, being one of those names early, before your competitors get there, is worth far more than the raw traffic count suggests. We treat this as positioning for where things are clearly heading. Not a tap you open for leads tomorrow.

Attribution is genuinely hard, and we won't fake it. When a homeowner calls, they don't say "ChatGPT sent me." We can measure how often you get named in AI answers. That's the deliverable, and it's real. We can't hand you a clean count of leads that came from AI. That data mostly doesn't exist yet. Anyone who hands you a precise count of AI leads made it up. The tools to track that aren't here.

This field moves in weeks, so we recount weekly — not once a quarter.

The sources AI draws on shift constantly. Over one documented stretch in 2025, one major platform's share of AI citations fell from around 60% to around 10% in roughly six weeks. A model update can reshuffle the whole thing overnight. A setup that had you visible last quarter can quietly stop working and never tell you.

So we don't run one audit and walk away. We re-check the ground truth week to week, not once a quarter. It's also, frankly, why we built our own measurement system instead of renting one. We wanted to watch your visibility at the speed the field actually moves and catch a shift the week it happens. Not the quarter after.

We measure two ways. When they disagree, you get "pending," not a guess.

A measurement system in a field this fast is most dangerous when it breaks quietly and keeps reporting confident numbers that are wrong. We built against exactly that. We measure AI answers at volume through the engine APIs, and through spot checks of what a real person sees in the consumer apps. When the two disagree past a tolerance, that's a flag. Our fast numbers might not match what a homeowner actually sees. So we dig in before we report anything.

If the system can't stand behind a number, you don't get that number. We'd rather tell you a result is pending than ship you something we don't trust. Saying nothing beats sounding sure and being wrong.

We also record what we saw and why we concluded it, not just the result. So when we tell you something changed, we can show you the basis for it. You don't have to take it on faith.

What's missing from this list is the point.

Here's what we won't claim.

• We won't give you an "AI rank position." It isn't a real thing.
• We won't promise you a set number of leads from AI. The data to back that up doesn't exist yet.
• We won't call the review floor, or any other figure here, an official algorithm rule. It's the best evidence we have, held as evidence.
• We won't pad your site with AI-generated filler. Thin, mass-produced content doesn't help. Lately it hurts.
• We won't tell you we can fix your AI visibility while your reviews sit below the floor. We'll tell you to fix the reviews first.