MarqueeOS / Marquee Visibility
Your buyers ask AI.
AI has never heard of you.
Marquee Visibility is answer engine optimization. We make your business the thing ChatGPT, Claude, Perplexity, and Google AI Overviews name when someone asks for what you sell.
- We measure first. Ten buyer questions, run against real answer engines, recorded verbatim. That's your baseline.
- Then we fix the reason. Identity, question coverage, quotable content, and entity corroboration. In that order.
- Then we re-run it monthly. Same questions, same instrument, so the trend line is honest.
- The proof metric is a citation, not a ranking. The first month you appear in an answer where you appear zero times today.
Most businesses are findable only if you already know the name.
Five quick questions, then pick a time.
What is answer engine optimization?
Answer engine optimization, or AEO, is the practice of making a business legible and quotable to AI answer engines so it gets named and cited when someone asks a buying question. Traditional SEO competes for a position in a list of ten blue links. AEO competes to be the source an AI reads, trusts, and repeats inside a single generated answer, where there's no page two and usually no more than five names. It's also called generative engine optimization, or GEO. Marquee Visibility is the MarqueeOS module that does this work, alongside Marquee Demand, which creates demand, and Marquee Pipeline, which captures and converts it.
We don't guess.
We ask the engines.
Every engagement opens with a scan. Ten questions a real buyer would type, split across three kinds, run against live answer engines, with every answer stored word for word. Here's the shape it usually takes.
Anonymised from a real baseline. Query wording, competitor names, and the client are withheld. Yours is run on your own category and your own city.
Search traffic may fade.
Being the answer doesn't.
A buyer used to compare ten links and pick one. Now they ask once and get a paragraph with three or four names in it. You're either in that paragraph or you aren't in the consideration set at all. There's no page two to come tenth on.
Numbers from one anonymised client baseline, 10 queries, replicated across four runs. Your own scan is the first deliverable, and it'll say something different.
Your SEO doesn't carry over.
Here's why.
An answer engine isn't ranking pages. It's assembling a paragraph, and to be in that paragraph your business has to be three things at once: identifiable as a specific entity, relevant to the exact question, and quotable in a sentence or two.
Most sites fail on the third one before the first two ever get tested. A page with a headline and forty words has nothing an engine can lift. A competitor with a plain FAQ answering the same question in ninety words gets quoted instead. We've watched a painting company get named for plaster work, not because they were better, but because they published a page that answered the question.
And when nothing on the open web corroborates who you are, the engine guesses. On one real scan the engine answering "who is this company" pulled from three unrelated businesses with similar names and a Wikipedia page for a venture capital firm, then hedged its own answer with "appears to be."
Five dimensions, scored out of 100.
The same audit runs on your site before we touch anything and after every month of work. It's deterministic, so the score moves only when the site does.
Bars show a representative starting profile. A first audit in this shape scores in the twenties, which is an F, and is completely normal for a site that was built to look good rather than to be read by a machine.
Four layers, in this order.
The order matters. Publishing answers before an engine knows who you are just donates content to a competitor's citation.
Identity
Structured data that states exactly what you are, where you operate, and how to reach you. LocalBusiness or the right industry subtype, full name address and phone, geo coordinates, service area, hours.
Questions
FAQ content and schema aimed at the questions your scan showed nobody winning. These need no domain authority, which makes them the cheapest citation available to you.
Answerable pages
Service and location pages written to be quoted, not skimmed. Real depth on the things you actually sell, in the places you actually serve.
Entity corroboration
The off-site signals that turn a claim into a fact. Profile consistency, sameAs links, and the listings an engine cross-checks before it repeats your name.
Three things, every month.
Not a login you'll forget. The work is done for you and the evidence arrives whether or not you go looking for it.
The scan
The same buyer questions, re-run against live answer engines, with every answer stored word for word.
- Named or absent, per question
- Which competitors got named instead
- Which page of yours got cited
- The answering model, on the record
The score
The five-dimension audit re-run on your site, so you can see the foundation move before the citations do.
- Reachable, Identity, Questions, Answerable, Entity
- Scored out of 100, deterministic
- Diffed against last month
- Every failing check named
The work
Shipped, not recommended. Schema, FAQ pages, service pages, and entity signals, live on your site.
- What shipped this month
- What ships next and why
- Written by a person, not a template
- One 30-minute review call
What happens, and when.
Nobody should promise you a citation in September. Engines re-crawl on their own schedule and the honest timeline is longer than the sales pitch you've probably heard.
The scan, the audit, and the identity layer shipped. You get the "before" snapshot that every later month is measured against.
FAQ pages targeting the queries where your scan found no local winner at all. This is the cheapest ground on the board and it's usually still empty.
Engines re-index. Realistically this is where the first category citations land, not before.
The scan re-runs monthly against the identical query set, so a change in the number means a change in you, not a change in the test.
What we can and can't measure.
Google AI Overviews and Copilot expose no API. Neither does the consumer ChatGPT app, which runs a different retrieval stack from the API of the same name. Anyone who tells you they're tracking your exact position inside a specific person's ChatGPT session is selling you something that doesn't exist.
What we measure is a close proxy: the same query set, run through real answer engines on a vendor API, recorded verbatim, and repeated identically every month. The instrument records which model answered and warns us if the vendor re-points it, because if November differs from October the first question is whether your site changed or the test did.
That's a real limitation and every tool in this category shares it. Most of them don't print it. You'll get the caveat in writing in every report, because a number you can't interrogate is worse than no number.
Seven questions to ask
an AI visibility vendor.
This category filled up fast and most of it is dashboards. A dashboard isn't the problem you have. You already know you're invisible. What you're buying is someone to fix it. Ask these before you sign anything, including with us.
We wrote these to be used against us too. If our answer to any of them ever stops being the one above, hold us to it.
Who this is for.
- Founder-led businesses doing $1M to $50M, roughly 5 to 50 people
- You sell something people research before they buy
- Your customers have started saying "I asked ChatGPT and it said..."
- You rank fine on Google for your own name and nowhere for what you actually do
- You'd rather own three questions completely than come tenth on one
Not a fit: pure impulse retail, businesses with no website worth fixing, and anyone who needs leads this month. This is a compounding asset, not a faucet. If you need the phone to ring now, start with Marquee Demand instead.
Questions.
Find out what AI says about you
when you're not in the room.
Five quick questions, then pick a time.