Here's what you'll get.
- ✓ A presence audit that goes beyond your website — we see what your overall presence looks like, and have AI tools describe what they see.
- ✓ 17 questions, built from what your website and presence actually say, run against the major AI engines to see how they respond — who gets recommended, who gets mentioned, and who doesn't show up at all. It can vary dramatically from AI engine to engine.
- ✓ Real-world fixes, included as part of the report.
You'll also see who your competitors are in that space — and why they were recommended over or under you across the different questions.
Here's exactly what's in it — scored across five components, 100 points:
Unbranded presence — do you come up when someone describes the problem and doesn't know your name?45 pts
Citation share — is your site a source the AI engines actually pull from?20 pts
Branded recognition — do the engines know you when someone names you?15 pts
Engine coverage — how many engines, so one lucky result can't carry the score10 pts
Entity accuracy — is what they say about you actually correct?10 pts
Bands: A 85+ · B 70+ · C 55+ · D 35+ · F under 35. The scoring is deterministic — the same captures always produce the same score, which is what makes your 90-day retest a comparison instead of an opinion.
The reports we produce
$195GEO St. Score™ ReportThe full baseline — twelve sections, your score and band, who owns the answers today, and the fixes ranked by leverage.
$294Report + Evidence togetherThe most common choice: the diagnosis and the raw proof behind it, in one purchase.
+$99Evidence Add-onEvery question and every engine's answer, verbatim — plus a competitor analysis and the authoritative sources the engines leaned on.
quotedMarket St. Radar™The standings board for a whole market: who gets recommended, and which questions nobody owns yet.
quotedEntity Probe™What the engines actually know about a business, and which sources taught them.
quotedCompetitor ReportsAny named competitor run through the same battery, side by side with you.
Example 1 — the GEO St. Score™ report
Here's what the top of the report looks like:
Sample Plumbing Co. · fictional example
38.0/100
Band D
Unbranded presence██░░░░░░░░ 9/45
Citation share██░░░░░░░░ 4/20
Branded recognition█████████░ 13/15
Engine coverage█████░░░░░ 5/10
Entity accuracy███████░░░ 7/10
Read that pattern: the engines know this business perfectly well when you name it, and almost never bring it up when you don't. That's the most common shape we see — retrievable, but not recommended.
And a line from each section of the report itself:
1 · Capture validity
Run ST-4471 · 68 of 68 cells captured · clean room VERIFIED · grounding confirmed on all four engines. Nothing below is an estimate.
3 · What the data says
Branded recognition 87% against unbranded presence 20%. When someone names this business, the engines know it. When they describe the problem instead, it is not in the room.
4 · Engine-by-engine
Perplexity named it on 3 of 17 questions. ChatGPT named it on 0. Same questions, same day — the engines ground on different sources, which is what makes the gap actionable.
5 · Who owns the answers today
Yelp, Angi and Thumbtack take 61% of citations on the unbranded questions. Other operators take 12% — and that is the most beatable tier there is.
6 · Who gets named instead of you
Three competitors appear on questions where this business does not. One of them is named on 11 of the 17.
8 · Where to attack first
“How much does repiping cost in Florida” — absent, and no incumbent owns it either. Absent plus fragmented is the highest-leverage page you can build this month.
9 & 10 · Site changes and real-world authority
The page build order that falls out of section 8, the entity fixes that make you one business instead of three, and the off-site moves — ranked by effort against payoff.
Twelve sections in all. Section 7 is the full verdict matrix — every question against every engine, so any single cell can be audited:
Live AI Test Results
| What a customer asks AI | Result |
|---|
| Best plumber near me in Tampa | ✗ Not mentioned |
| Emergency water heater repair Tampa | ◎ Mentioned |
| How much does repiping cost in Florida | ✗ Not mentioned |
| Plumber with best reviews South Tampa | ✗ Not m— |
The rest of this report is where your customers went.
Every question is run blind. We test each one from a fresh AI session with no account history and no personal signals — the same way an AI helping a stranger would see you. If you run these questions yourself from your own logged-in AI accounts, you may get noticeably better answers — not because that's what's publicly findable, but because that AI may already remember your business from your own past conversations with it. Our report shows what a new customer's AI actually sees.
Want to dig into the data yourself? Add the full text of every query for $99. It comes with two additional analyses: a competitor analysis, and a look at the authoritative sources the AI engines used to answer the questions — separate from the analysis that's already part of the GEO St. Score™ report itself.
Example 2 — what AI expects from your content
The score tells you where you stand. This tells you how to write so an engine will quote you. Every report ships with the writing method behind its fixes — the habits that decide whether an AI engine lifts your sentence or skips past it. Here is one of them in full.
Coastal Roofing Co. · a different fictional example
Technique 1 — Front-load the direct answer
AI engines lift the first sentence as the quotable answer. If sentence one is a hook or a story, the engine quotes something vague — or skips the passage entirely.
The question a customer actually asks: “Do I need a new roof or just a repair?”
Before — Most homeowners in Pinellas County don’t think about their roof until a storm forces the issue.
After — You need a full replacement if shingles are curling across more than a third of the roof, or if the decking feels soft underfoot; below that, a repair usually holds. Most homeowners in Pinellas County don’t think about it until a storm forces the issue.
Same content. Same voice. The answer just moved to the front — and now it stands on its own when an engine lifts it out of context.
Technique 2 — Make every answer self-contained
An engine will pull one paragraph out of context, and the reader may never see the question it answered. So an answer cannot lean on “this,” “it,” or “that number” without restating what they refer to.
Before — It’s what happens when the underlayment gives out before the shingles do.
After — Premature underlayment failure is what happens when the waterproof layer beneath your shingles br—
There are several more — and your report names the ones your pages are missing.
Get your GEO St. Score™ →
