How to make money with AI visibility if you run a small agency

You make money on this by selling a fixed-scope AI visibility audit to an ecommerce client, out of a budget they already have, and turning it into a monthly read once they've seen the gap. That's the whole play. Small, fast, no systems access, and it converts because the client sees a competitor named where they aren't.

What follows is who to sell it to, what to charge, and the one thing you must never put in writing.

Because you've had the other version of this meeting. The client brings up ChatGPT, you both agree it matters, everybody nods, and nothing happens. No brief, no budget. Three months later they raise it again like it's the first time anyone's mentioned it.

Everybody's asking. Almost nobody's buying. Half of marketers now watch what the models say about them, twice as many as a year ago, and at your end of the market it still isn't turning into work. The reason is that they're waiting for a budget to appear and so are you. It isn't going to. You have to go and take this out of one that already exists.

Nobody signs off a line item called GEO. They've never bought one, they can't benchmark the price, and it sounds like a thing you invented on the way to the meeting.

They will buy a one-off audit. Fixed price, fixed scope, delivered in a week. You run their real buyer questions through the engines, you show them who gets named, and you show them the pages the engines read to decide. No access to their systems, no data processing agreement, no security review. Nothing to approve except a number.

That's the wedge, and it's the whole reason this is sellable to a small client at all. You're not asking them to believe in a category. You're asking them to look at one report.

The retainer comes after, and it sells itself, because a single run is a photograph and everybody who sees it immediately asks whether it's moving. Price the audit so you'd be happy doing only that, and treat every retainer as upside.

If you sell this to one segment, sell it here, and it isn't close. Three reasons and they stack.

The money's already committed and it hurts. A DTC brand under a million in revenue is putting a quarter to a third of everything it makes into being found. A quarter of the business, out the door, every month, so a stranger sees the product. They know exactly what that feels like and they resent it, deeply, the way ecom operators always have.

One person signs. A founder or a growth lead. No committee, no legal review, no eight weeks.

And their customers are genuinely shopping this way now — somewhere between four and seven in ten, depending on whose definition you use, and I'd quote the range rather than pick the flattering end. The direction isn't in doubt either way.

So the pitch is one sentence and it never mentions a new category: you're spending a quarter of your revenue to be found, and there's a place where buyers are choosing between you and three competitors that you can't see at all. Say that, show the report, stop talking.

Start with a conversation, not an audit. Sit down with the owner and explain what's actually changing, because most of them haven't thought about it once.

People are asking AI about products and services now. On the laptop, on the phone, and more and more just speaking to it. What's the best Thai place near me. Can you recommend a plumber in Hitchin. Is there a decent dentist round here. It's the same habit they built typing into a search bar, moving across to an assistant — and what comes back is two or three business names.

Their client's name is either in that answer or it isn't.

That's the entire pitch and it doesn't need a deck. You've spent years getting this business visible in search and on the map, because that's where people were asking. This is the next place they're asking. It's early, it's trending up, and like everything else you've done for them it takes steady work rather than one fix — the map listing didn't happen in a week either.

Then show them. An AI visibility audit is what turns the conversation into a number: which engines name them, which name a competitor instead, the exact questions where they're missing, and the pages the engines read to make up their mind.

If they want to see one before they commit, we published ten audits of national brands. Here's an example. Not a local business, so the categories won't look like theirs, but the shape is identical and it's easier to understand a thing you can scroll than a thing you're describing. Show them that, then run theirs.

And don't ask for new money. Go in with their audit already done, put the competitor's name on the screen, and ask to move a few hundred a month off something that isn't working. That's a conversation they'll have. "We should look at GEO" is one they'll defer.

The most useful thing about the enterprise end isn't that you'll sell to it this quarter. It's that it tells you this is real.

National brands are standing up dedicated AI visibility budgets with committees attached and funding them properly. Seven to twenty-one thousand a month is the going rate for the search programme alone up there, before anybody adds this on top. The companies with the most to lose and the best data are moving first, which is usually how you tell a channel is arriving rather than trending.

Use that in the room. When an owner asks whether this is a fad, the answer isn't a statistic — it's that the biggest brands in their category already have somebody whose job this is.

Selling to them yourself is slower. Big companies buy through procurement and that's its own project, so the way in is a small pilot rather than a retainer pitch — the same fixed-scope audit, small enough to go on a card. Worth starting. Just don't build this quarter on it.

This is the one that ends retainers, so put it in the scope document before it ever reaches an invoice.

There is no position to promise. A Google ranking is a place on a list you can go and look at — type the search, count down the page, there you are. An AI answer isn't a list at all. It's written fresh each time from a stack of pages the engine picked and read on the buyer's behalf, and nobody gets to see that stack. A screenshot of one good answer isn't a result, it's a moment.

What you sell instead is the thing that does hold still. How often the brand gets named across fifty real buyer questions on every engine, measured against its competitors by name, and what moved since last month. That's a number you can defend at renewal and it's yours to move.

Underneath it sits the part clients have never been shown before: the reading list. The actual pages the engines went and read to reach that answer — which ones carry your client, which ones hand the answer to a rival, and which ones won't even open. That's your work order for the next quarter, and it's the reason the monthly exists.

So price the audit as a fixed fee, and price the monthly as a reading rather than a ranking.

Take the ecommerce account on your books with the most to lose. Run their real buyer questions through the assistants and see who comes back.

If three competitors get named and they don't, you've got your first sale — and it isn't a sale about GEO. It's about the quarter of their revenue they already spend on being found, and the part of it that's going nowhere.

Twenty minutes. Better use of the afternoon than waiting on a budget being written three tiers above your client list. And for what it's worth, the numbers here are sourced but the play is mine — I'd rather you tested it on one account than took my word for it.

How does a small agency actually make money on AI visibility?

Sell a fixed-scope audit first — the client's real buyer questions run through the engines, showing who gets named and which third-party pages decided it. Fixed price, delivered in a week, no systems access and nothing for legal to review. The monthly retainer follows on its own, because a single run is a snapshot and every client who sees one asks whether it's moving.

Which clients actually pay for GEO work?

Ecommerce and DTC brands are the best odds today: budget already committed to being found, a single decision-maker, and customers visibly using AI assistants to shop. SMBs will buy it out of an existing programme rather than new money. Enterprises have the largest budgets and the slowest door.

Has budget really shifted from SEO to AI search?

At the top of the market, yes — national brands are funding dedicated AI visibility work and running committees for it. What no source shows is the market-wide figure: how much has moved, from where, across how many companies. Marketer attention has clearly moved (Fractl found 49% now monitor LLM impact on their brand, up from 22% a year before), but attention isn't a budget line, and anyone quoting you a reallocation percentage for the whole market should be asked where it came from.

Why is enterprise so hard for a small agency?

Procurement. An MSA, a DPA, insurance certificates, a security questionnaire and often a SOC 2 report, routed through legal, IT and finance. It's clearable but it's unpaid project work that runs eight weeks minimum. Enterprise SEO programmes run $7,000 to $21,000+ a month and the combined stack is put at $80,000 to $250,000 a year, so it's worth reaching — via a small pilot engagement, not a retainer pitch.

Can I promise a client a position in AI answers?

No, because there isn't one. A Google ranking is a place on a list you can go and count down; an AI answer is composed fresh each time from a set of pages the engine chose and read on the buyer's behalf. There is no slot to occupy. What you can report is share of voice — how often the brand is named across many real buyer questions on each engine, measured against named competitors — and how that moves between runs. StyleForge's field study found that measurement stable enough to report on: brands were mentioned in 69.0% of answers on the first run and 68.3% on a re-run the following day.

What should I sell first?

A fixed-scope audit for an ecommerce brand in a category nobody's competing in yet. Fast, cheap to deliver, no systems access — and a clean early result is what earns the retainer conversation.

• How to sell GEO services — scoping and pricing the engagement once you've got the meeting.
• Your client asked ChatGPT about their own industry — the diagnostic, in the order I'd work it.
• GEO vs SEO: what the hell is the difference? — what the engines actually read before they answer.
• The best channel for winning SEO clients — what the survey data says about where business comes from.