The first time an agency owner asked me how to start an AEO practice was over an espresso in Lisbon, early February. He had ten SEO retainers on the books, each one bleeding traffic. He stirred his coffee and asked the question I have since been asked forty times. "If I were starting over tomorrow, knowing what we both know about AI search, what would I build?"
The honest answer is that the window is open and it is moving fast. Organic click-through drops 61% when an AI Overview triggers, and AI Overviews now trigger on between a quarter and nearly half of tracked queries. Sixty-three percent of agencies changed their KPIs last year because of the AI shift, and almost a quarter of them have already lost clients to it. That is the macro picture. The micro picture is the agency owner in front of you, staring at a P&L that no longer adds up and trying to figure out what comes next.
What follows is the essay I wish I had handed to my Lisbon friend that morning. It is not a step list. It is closer to a long letter from a founder who has watched dozens of agencies pivot into AEO and has formed strong opinions about what works, what does not, and where almost everyone gets stuck.
The shape of the opportunity
Begin with the economics of the broken thing. An SEO agency made money the same way for fifteen years. You sold a retainer for three to twelve thousand dollars a month. You promised rankings, traffic, and eventually pipeline. Your client trusted the rising green lines because organic traffic and rankings actually correlated with revenue.
They no longer do. When Google answers the query on the results page, the click never happens. Your ranking is intact. Your traffic is not. The client sees a cliff in organic sessions and asks why. You can explain the SERP shape change. None of that pays for the retainer. The client is buying outcomes, not explanations, and the outcome they used to buy has been quietly de-listed from the menu.
Meanwhile, on the other side of that same shift, an entirely different category of work is suddenly worth more. Being cited inside an AI Overview is worth roughly 35% more organic clicks and 91% more paid clicks than being absent. AI-sourced visitors spend 68% more time on the site than ordinary organic visitors, and they convert at roughly three times the rate. The pie shrank. The slice you actually want got larger, denser, and far harder to win. That is why the agencies that figure out how to deliver it are pricing higher than they ever priced for SEO.
"You did your job. Google changed the rules. An entirely new ranking system has appeared, and somebody is going to be the agency that helps everyone show up inside it."
The historical analogue I keep coming back to is Amazon FBA around 2014. Jungle Scout did not invent FBA. It just owned the phrase "how to start an Amazon business" before anyone else did, sold the picks and shovels to a generation of sellers, and built a hundred-million-dollar business off a category nobody could yet name. AEO is in the same window now. Somebody will be the platform every new agency runs on, and somebody else will be the agency that defined the category in their vertical. This essay is mostly about the second one.
What an AEO agency actually does, when you take it apart
I want to resist giving you a tidy seven-point list, because an AEO agency has a particular operational shape and you only learn that shape by spending a few months inside one. Still, you can describe the layers, and once you see them you cannot un-see them.
Finding the people who will pay you
Lead generation in this category is not keyword research and not pure list-buying. The agencies that pull it off well start by picking one vertical — SaaS founders, US personal injury law, multi-location dental, mid-market HVAC, Shopify brands doing seven to twenty million in revenue — and build a list of one hundred to two hundred companies that fit a clear ideal customer profile. The hook is not "do you want SEO" but "did you know your brand is invisible inside ChatGPT for the three queries your buyers actually run when they are about to choose someone?" That is a different conversation, and it opens doors the older sales motion no longer opens.
Turning the list into discovery calls
Outreach is mostly cold email, with LinkedIn touches as a secondary channel. The agencies that hit reply rates above the baseline almost always do the same two things. They run a quick free citation audit on the prospect before sending the first message, so the email opens with something specific the prospect did not know about themselves. And they keep the message short — three or four sentences, one concrete fact, one specific ask. The baseline conversion from cold email to a fifteen-minute call hovers around two to three percent. Brutal arithmetic at twenty prospects. Workable at two hundred.
The pitch where the deal is actually won
A discovery call for an AEO retainer is closer to a diagnostic than a sales call. You walk the prospect through their citation score across the five engines that matter — ChatGPT, Perplexity, Gemini, Claude, and the citation set behind Google AI Overviews — and you let the numbers do the persuasion. Most prospects have never seen this view of their brand. They have a vague sense that AI is happening to them and a precise sense that their traffic has softened, but they have never seen the two stitched together. When you draw that picture, the question stops being "should we do AEO" and becomes "what tier of your retainer should we sign up for."
Measurement, daily and dull
Once a client is on the books, the operational heart of the relationship is daily citation tracking. Every morning, against the ten to thirty tracked queries you negotiated, your platform sweeps the five engines and records the answer. Was the client cited? In what position? Which sources did the model lean on? Did a competitor displace them overnight? This is dull, load-bearing work. It is what justifies next month's invoice and what lets you walk into a quarterly review with hard numbers instead of stories.
Content and the artefacts the client actually sees
The content half of an AEO retainer is less about volume than about shape. Pages organised around fifty-to-one-hundred-fifty word self-contained chunks, with question-shaped subheadings, FAQ schema, concrete statistics, and source attribution, get cited far more often than the long flowing essays traditional SEO production produced. Your client, meanwhile, does not consume the daily sweep or the chunked content. Your client consumes two artefacts: the monthly white-labeled report and the quarterly business review. Most agencies under-invest in those artefacts and over-invest in the operational layer. The artefacts are how the client decides whether to renew.
The economics, told plainly
Pricing is where new agency owners freeze. They have heard that AEO is the next big thing, they believe it intellectually, and they still find themselves on the discovery call quoting fifteen hundred dollars a month because that is what their last freelance gig paid them. Do not do this. The market has moved past it.
The agencies winning right now run a three-tier menu and refuse to discount below the floor. Foundation, at roughly three thousand a month, covers daily citation tracking, ten to fifteen queries, monthly reporting, and one strategy call. Growth, at five thousand, adds twenty to thirty queries, competitive intelligence, content optimisation, and a second strategy touch. Authority, at eight thousand and up, is for clients who want a quarterly business review with C-suite-grade material, forty-plus queries, four pieces of native content per month, and a white-labeled micro-site. Three named choices is a much easier conversation than an open-ended quote.
The unit economics underneath are unromantic and reassuring. A platform subscription that covers tracking, content, and reporting for a single brand runs between seventy-five and one hundred and fifty dollars a month. Your labour cost on a Foundation client, assuming a fully loaded eight-thousand monthly salary for whoever delivers, sits between three and four thousand. The gross margin on a three-thousand retainer lands between forty and fifty percent. Move that client to Growth and the labour does not grow linearly; the margin widens. The first three clients are usually break-even because you are over-delivering for case studies, and that is fine. By client five, the agency is comfortably profitable.
"Price your first three retainers at the floor to lock in case studies. Then raise everyone afterwards. Do not discount below the floor; it teaches a market that your category is cheap."
A prospect will tell you they cannot afford three thousand. You will be tempted to drop the price. Drop the scope instead. Cut the tracked query count in half, defer the content optimisation, push the first strategy call out by a month. The price signals the category. If you sell AEO for fifteen hundred, the next agency in your vertical will not be able to sell it for three thousand, and the category becomes the next cheap commodity instead of the next premium service.
On the temptation to assemble your own stack
Almost every founder wants to build their own stack first. The instinct is rational. Apollo for prospecting, Lemlist for outreach, Frase for content, Pipedrive for the CRM, Webflow for client micro-sites, Profound for citation tracking, and AgencyAnalytics for white-labelled reports. At list prices, you are at roughly six hundred and ninety dollars a month before onboarding a single client. Add a dedicated citation seat per brand and you are closer to a thousand. That is a real number, but it is not the number that matters.
The number that matters is the time spent stitching these tools together. Five to ten hours a month of glue work that does not show up in any client report. Zapier flows that break when an API endpoint changes. Lead enrichment that arrives in Apollo but never makes it to the outreach tool. Citation snapshots that have to be exported, formatted, branded, and pasted into the report. For an agency owner whose only scarce resource is their own attention, that hidden cost is more expensive than the cash difference. I have watched founders run a multi-tool stack for six months, land client five, and quietly migrate the entire operation onto a single platform in a weekend because they finally hit the moment where coordination cost exceeded consolidation cost.
The competitive map is roughly three categories. There are the pure-play AEO trackers — Profound, Peec, Otterly — which solve the measurement problem cleanly and leave the rest of the agency operation to you. There are the SEO incumbents bolting on AEO modules. And there is a small but growing set of agency-first platforms designed to handle prospecting, outreach, content, tracking, and reporting as one surface. The choice is less about features than about who you want to be in two years.
The first thirty days, as I would do them now
If I were starting an AEO agency next Monday with nothing but a laptop, here is how the first month would go. I am writing this in prose deliberately because the moment you reduce it to a checklist you start to believe the work is linear. It is not. Most of the early weeks are a tangled mess of doing four things at once.
The first week is about getting the agency real enough to charge money. Register the entity within a day. Buy a domain that suggests AEO without sounding like a parody. Open the business banking, knowing it will take five business days to clear. Pick a single vertical and write a one-paragraph ideal customer profile so specific it would embarrass you to share publicly. "Multi-location dental practices in the southeastern United States with annual revenue between three and twelve million, currently spending more than five thousand a month on a digital agency, with at least three practitioners." That paragraph is the entire first week of strategy.
The second week is the list and the sequence. Build a hundred and fifty prospects against that profile. Write the first cold email by hand, with a specific factual hook. "I ran your practice through five AI search engines this morning, and on the seven queries I tested, you appear on one of them." Three follow-ups, two to three days apart, each one shorter. By the end of the second week, expect five to fifteen replies, of which two or three are genuine discovery-call requests.
The third week is the audit and the calls. For every warm lead, run a proper citation audit before the conversation — not just the public score but the full query matrix, with the specific sources their competitors get cited through. Fifteen-minute calls, three slides, one citation matrix. The goal of the first call is not to close. The goal is to make the prospect feel seen, show them a number they have never tracked, and make them care enough about it to want it to change.
The fourth week is the proposal and the onboarding. Pick the warmest lead and send a one-page proposal. Foundation tier, three thousand a month, three-month minimum, scope stated clearly. Use PandaDoc or DocuSign so the signature is one click. When they sign, set up their tracked queries, send the first daily citation report on day one, calendar the first monthly check-in. You now have a paying client and a real case study compounding for you.
On vertical choice, the only strategic decision that matters early
Almost every failure mode I have watched a new AEO agency walk into has a single root cause. They tried to be generalist. They pitched "AEO for B2B companies" and the result was a cold email that landed like every other vendor pitch, a website that could have belonged to anyone, and a case study that did not translate from one client to the next. The generalist agency feels rational because it preserves optionality, and it slowly kills the founder by stretching every function thin.
A vertical-first agency makes every part of the business cheaper. Cold email gets sharper because you can reference the specific way buyers in that vertical search. Discovery decks become reusable. Case studies compound because your second client cares about your first client's outcome. Content production cycles faster because you build a domain library the second time around. Pricing benchmarks itself. None of this is true the first time you sell, and all of it is true by the time you sell your fifth.
The right vertical is the one where you have at least one of three things — domain expertise, an existing network, or a buying motion you have personally felt. The verticals that win for outsider founders in this category are typically the high-margin services where local intent is strong and buyers have started using AI tools in their decision: medical practices, regulated professional services, mid-market B2B SaaS, ecommerce in considered categories. The verticals that struggle are commodity B2C and low-AOV ecommerce, where the unit economics will not support the retainer.
The honest list of ways this can still fail
I want to spend the last few paragraphs on failure modes because most essays about new agency categories skip them, and the result is a generation of founders surprised by the same small set of problems.
The first failure mode is under-pricing. You will be tempted, on your first three calls, to win by being cheaper than competition that does not yet exist. Do not. Win by being more specific. The clients you attract with low pricing are exactly the clients who churn the moment they decide to bring it in-house.
The second failure mode is over-promising. The temptation is to guarantee a specific citation lift inside ninety days. Resist. Citation movement is real and sometimes very fast, but it depends on the client's content inventory, their existing domain authority, and the competitive density of their vertical. Promise process — daily measurement, weekly content, monthly reporting, quarterly business review — not outcomes you do not control.
The third failure mode is hiring too early. The first three clients are deliverable solo. The fourth and fifth are deliverable solo if you are disciplined. Somewhere between client five and client seven, you will need help — usually a part-time operations person and a freelance content editor. Bring them in any earlier and you spend the runway you should have been using to find clients on payroll. Bring them in any later and you start dropping balls in front of clients paying you to keep them in the air.
The fourth failure mode, and the most underrated, is treating AEO as a different product instead of a different layer. The agencies now doing fifty thousand in monthly recurring revenue did not sell AEO as a replacement for the rest of the client's marketing. They sold it as the new visibility layer that sits underneath everything else. Position yourself as the addition, not the replacement, and you will close more deals.
Where this leaves you
Back to Lisbon, and the question my friend asked me over coffee. If you were starting an AEO agency next Monday, knowing what we both know, what exactly would you build? You would pick a vertical you understand. You would build a list of a hundred and fifty prospects and run a citation audit on each. You would write a sharp cold email with a specific factual hook. You would book discovery calls and let the citation matrix do the persuasion. You would price at three named tiers and refuse to discount below the floor. You would invest unreasonably in your monthly reporting because it is what the client actually consumes. You would resist the urge to assemble seven tools and instead pick a single platform that lets you spend your scarce hours on clients, not glue work. You would treat your first three clients as case studies disguised as contracts.
The macro numbers are still moving in your direction. The AEO-as-a-Service category is growing at roughly 71% a year and is forecast to clear $340 million by 2027, up from a small fraction of that two years ago. The buyer side is even cleaner: 94% of CMOs and digital leaders plan to increase AEO spend next year. The window will close, eventually, the way every new category closes — when the incumbents arrive, pricing settles, and the playbook becomes common knowledge. By then, the agencies that started this quarter will have their first dozen retainers, their first three vertical case studies, and the kind of inbound that only comes from being early enough to be remembered.
The Lisbon conversation ended with the same question framed differently. "So what is the actual first step on Monday morning?" The actual first step is the most boring one. Pick the vertical. Write the paragraph that describes the ideal client so precisely it would embarrass you. Build the list. Run the audit. Send the email. The rest follows from that single decision, made tightly and held to.
If you are building this agency right now, GenPicked's Growth plan is the operating system for it. Start your 14-day free trial — Growth plan free for 14 days, five AI engines, full agency dashboard.