Structure
Written to be extracted
Content shaped so a model can lift a clear, correct, quotable answer out of it — which is a different job from being crawlable.
Kellstar Corp — GEO
Nobody scrolls when the answer is already there.
More buyers now ask an assistant instead of searching. It answers in one paragraph and names a handful of sources. GEO is the work of being one of them — ongoing, measured by actually asking the assistants, not by a proxy metric standing in for it.
$200 Basic (upkeep optional) · $578 Intermediate · $1,679 Enterprise · month to month
01 A different question
Not a rename
of SEO.
They overlap, but they aren't the same measurement — and a lot of what's currently sold as GEO is SEO with a new label on the invoice. The distinction is simple enough to state in two sentences.
SEO asks
"Does this page rank?"
A position in a list of links, competing for a click. Measured by where you sit on the results page.
GEO asks
"Does the answer mention us at all?"
A named source inside a written answer, where there is no list and usually no click. Measured by asking the question and reading what comes back.
02 What the month actually buys
Three things,
done continuously.
Trust signals
What makes a source citable
The factors that make a page worth naming as a source. They aren't identical to the factors that make it rank, which is why doing SEO alone doesn't get you here.
Measurement
Checked against real answers
Tracked by actually asking assistants the questions your buyers ask and recording whether you're named. No proxy metric standing in for the thing itself.
03 Free readiness scan
What we check
before selling
you anything.
Send us a URL and we run it. The scan reads your robots.txt, your markup and the HTML you actually serve, scores eight things out of 100, ranks you against competitors you name, and comes back as a report with the work in the order it should happen. Free, no call required, usually same day.
What it scoresWeight
25pts
AI crawler access
Ten assistant crawlers, each checked by name against your robots.txt. Weighted highest because nothing below it matters if the crawler never arrives.
20pts
Structured data
Whether an assistant can read who you are, where you operate and what you sell as facts — or has to infer all three from prose and get it wrong.
15pts
Answer-shaped content
Question-led sections are the thing that actually gets lifted into an answer. Most sites are written as brochures and have none at all.
15pts
Entity clarity
Name, address, phone, page title and description present as plain text — the facts a model corroborates you against before it decides you're the one to name.
10pts
Readable without JavaScript
Most assistant crawlers don't execute it. If your copy only appears after a framework boots, they fetch a nearly empty page and leave.
5pts
Content depth
Enough substance on the page to be worth quoting.
5pts
llms.txt
An emerging convention — a plain-language guide written for assistants. Cheap to add, almost nobody has one.
5pts
Sitemap
Present and reachable at the usual path.
Crawlers checked by name10
GPTBot ChatGPT
OAI-SearchBot ChatGPT Search
ChatGPT-User ChatGPT visiting on request
ClaudeBot Claude
anthropic-ai Claude, legacy token
PerplexityBot Perplexity
Google-Extended AI Overviews / Gemini
Applebot-Extended Apple Intelligence
CCBot Common Crawl
meta-externalagent Meta AI
The limit
Does it tell you whether ChatGPT names you today?
No — and anyone selling you a number for that is inventing it. The answers move between runs, between users and between models, and a figure produced in a script here would be worse than none.
The scan measures readiness. Whether you're currently named is something we check the honest way: we open an assistant with you on the call and type the five questions your buyers would ask. You watch what comes back. That takes four minutes and it's the only version of that answer worth having.
Proof
Have you run it on yourselves?
93
/ 100
this page · 8 Sep 2026
Full crawler access, structured data, question-led sections,
llms.txt, sitemap. The seven points we're missing
are a published phone number and street address — a one-person
company that works by email doesn't have a shopfront, and we'd
rather show you the gap than fake the fix.
Run the same checks on us yourself: robots.txt, llms.txt, sitemap.xml.
Week 1 · only if found
Unblock the assistant crawlers
Skipped when nothing is blocked. When something is, it jumps the queue — everything else is wasted until it's fixed.
Week 1–3
Structured data for the business and its services
Who, where, what and how much, stated as machine-readable fact instead of left to inference.
Week 3–5
Rewrite the key pages as answers
The buying questions, answered in a form that can be lifted and quoted without editing.
Week 5–8
Entity and citation groundwork
Consistent facts across the site and the places assistants corroborate against — what decides which of several candidates gets named.
Week 8–12
Measure, then widen
The same questions re-asked and logged monthly. Whatever still isn't naming you tells us what to build next.
Every scan ends at the same recommendation — Intermediate, three months upfront — and you're being told that before you request one. What the scan changes is the order of the work, not the destination. $1,647
04 Tiers
Three levels.
All month to month.
Basic is deliberately different from the other two: the setup pass is most of the value, so after the first month the upkeep is optional. Take the groundwork and stop, or keep it maintained — both are real choices rather than one being the trap.
Basic
$200
first month · upkeep optional after
For getting the groundwork done properly once, without committing to anything ongoing.
- Full setup pass — structure, extractability, trust signals
- Baseline measurement against real assistant answers
- Monthly upkeep optional — continue only if you want to
- No ongoing content work
Intermediate
$578
per month
For a business that wants to be named across the questions its buyers actually ask.
- Everything in Basic, maintained
- Ongoing content across a wider set of questions
- Tracked monthly by actually asking the assistants
- Monthly review call
Enterprise
$1,679
per month
For a category where being the cited source is the channel, not a side bet.
- Everything in Intermediate
- Competitive monitoring — who else is being named
- Broad question coverage across assistants
- Priority turnaround and a direct line
Month to month, from the first month. No contract, no minimum term and no setup fee at any tier — after the first month, continuing is entirely your call. On Basic that's the whole point: the upkeep is optional, and stopping doesn't undo the work.
05 Start
Pick a tier.
Cancel any time.
No contract, no setup fee, no minimum term. Tell us the site and the questions you'd want to be the answer to — we'll confirm within one business day, including whether this is worth doing yet for your market.
06 Straight answers
Before you
start.
Eight factors, scored out of 100: AI crawler access (25 points — ten assistant crawlers checked by name against your robots.txt), structured data (20), answer-shaped content (15), entity clarity (15), whether the content is readable without JavaScript (10), content depth (5), llms.txt (5) and sitemap (5). Access is weighted highest because nothing else matters if the crawler never arrives. The full breakdown is above.
No. It measures readiness to be cited, not current citation. Assistant answers vary between runs, between users and between models, so a number generated in a script would be misleading — and you'd be right to distrust it. Whether you're named right now gets checked the honest way: on a call, asking assistants the questions your buyers would ask, with you watching what comes back.
For some businesses, yes — and we'll tell you that rather than take the subscription. It's worth doing when your buyers are plausibly asking assistants about your category already. If they aren't yet, SEO is the better use of the same money and we'd rather say so now than have you cancel in month three.
No, and neither can anyone else. These systems change without notice and nobody controls their output. What's guaranteed is that the work is real, the measurement is real, and you'll be told plainly whether it's moving — including when it isn't.
Most businesses should do SEO first — the fundamentals overlap and search is still where more buying starts. GEO makes most sense alongside it, or on its own if your market has genuinely moved to asking assistants. Not sure which? A consultation is $250 and will answer it properly.
Everything done stays done and stays yours. Cancelling stops future work; it doesn't remove what's already on your site.