What decides it: who rewrites the pages next month?

What decides it is who rewrites the pages next month, and the month after that. Every approach to AI visibility splits into two jobs. One is measuring where you stand: which engines cite you, on which prompts, and against which competitors. The other is doing the work, which means rebuilding pages so an engine can read them, lift a self-contained answer out of them, and treat the claim as credible. A tool covers the first job. In-house can cover both if you build the skill. An agency covers both as a service. Almost every disappointing AI visibility outcome traces back to confusing the two, because a dashboard that measures beautifully still changes nothing on its own.

The reason the second job never finishes is that citations do not hold. Trakkr Research ran identical prompts daily across 10,991 tracked brands and found that 73.5% of cited URLs appeared exactly once and never came back, with the median brand falling to half its peak citation count in 31 days, in its study The Half-Life of AI Citations, checked 25 September 2026. A page you fixed in March is not a page that is still winning in June. That is what turns AEO from a project into a standing job, and it is the fact that should decide which of the four models below you pick. Our own measurement methodology sets out the re-measurement cadence that follows from it.

Which model covers what?

No single model covers everything, so choosing one is really choosing which gap you can live with. Four criteria decide it, and they are worth stating before the options rather than after them. Who does the page rewriting, because that is the work that moves the number. How much of your own time the model eats once it is running, which is the cost buyers underestimate most often. How quickly you see a first result, which matters if you have to justify the spend internally before the year is out. And what happens when the person who owns it leaves, because AEO knowledge tends to sit in one head. The table sets the four models against those criteria.

Criterion	AEO tool	In-house	Agency	Hybrid
Who rewrites the pages	Nobody. It measures	Your team	The agency	Your team, on someone else's priorities
Your time once it is running	Low to read, high to act	Highest of the four	Lowest: review and approvals	Writing time, without the research time
Speed to a first result	Fast to measure, slow to fix	Slowest, then it compounds	Fastest end to end	Slower than an agency, faster than learning alone
Learning curve you take on	Some, to read the data well	All of it, and it keeps moving	None required	Execution only
Cost shape	Low monthly fee	Salary plus ramp time	Higher fee, the work included	Salary plus a smaller retainer
If the owner leaves	Data survives, nothing was being done anyway	The capability leaves with them	Nothing changes for you	Execution stalls, the strategy record stays
Who it is for	Teams with someone already doing the fixing	Companies where AI visibility is a permanent core function	Companies that want the result without building the skill	Teams with writing capacity but no appetite for the research
Main risk	Data nobody acts on	Pulling a person onto a moving target	Picking a repackaged SEO shop	A quarterly strategy deck with no shipping behind it

When is a tool enough on its own?

A tool is enough on its own when somebody, in-house or freelance, is already rewriting pages and only lacks the data to aim at. The tool shows you the prompts where a competitor is named and you are not, and which engine is the weak one. If you have the capacity to rebuild those pages afterwards, buying measurement and nothing else is the cheapest path by a wide margin.

The failure mode is well documented in every team that has tried it: you buy the dashboard, the dashboard is good, and nobody is ever free to act on it. Six months later you can describe your invisibility in detail and it has not changed. Before you buy, name the person who will do the rewriting and check what else is on their plate that quarter. If you cannot name them, you are not buying a tool, you are buying a report. What good measurement should capture in the first place is set out in our guide to tracking AI visibility across the four engines.

When does building AEO in-house make sense?

Building AEO in-house makes sense when AI visibility is going to be a permanent function rather than a campaign, and when one person can give it a few hours every week without it being the first thing dropped in a busy month. The discipline is about two years old. Engine backends change, citation behavior shifts, and the owner is signing up to keep relearning it rather than to learn it once. Where this is central to how you get found, that investment compounds and the knowledge stays in the building.

The common mistake is assuming an existing SEO hire already covers it. Ranking skill transfers less than you would expect. ConvertMate's GEO Benchmark 2026, built on more than 12,500 queries, found that 83% of AI Overview citations point to pages outside the organic top 10, per its GEO Benchmark Study 2026, checked 25 September 2026. The position work your SEO team has been doing is not what decides whether an engine quotes you. Structure, extractability and the language a buyer actually types are doing that instead, and somebody has to learn the difference. In-house means that somebody is on your payroll, learning on your time.

When is an agency the right call?

An agency is the right call when you want the result without building the capability, and when the months an in-house owner would spend getting good are months you would rather not lose. Measurement and the page work arrive together, you review and approve rather than research and write, and the knowledge problem stays somebody else's. The fee is higher than a tool because the work is included in it rather than sold separately.

The risk here is selection rather than execution. The category is new enough that an SEO shop can add AEO to its services page and sell the same retainer, and the buyer usually cannot tell in a first call. Ask which engine each of their clients is weakest on, which is a question that only has an answer if somebody measured the engines separately. Ask to see a dated AI answer that names a client. Our guide to choosing an AEO agency sets out the rest of the tests and what to write into the contract before you sign it.

What does the hybrid model look like, and when does it beat both?

The hybrid model keeps execution in-house and buys the thinking in from outside on a fixed cadence, usually a quarterly strategy session sitting on top of a standing measurement feed. Your team writes and ships the pages, because they know the product and they are already in the CMS. The outside party sets the priorities, reads what the engines have started doing differently, and says what changed since last quarter. Most comparisons leave this option out, and it is the one a good number of teams land on in their second year, once the novelty has worn off and the work has settled into a routine.

It beats a pure in-house build when your people can write perfectly well but nobody wants to own the research problem of tracking four engines that keep moving. It beats a pure agency when you have content capacity sitting idle and paying somebody else to write your pages feels like buying what you already have. It is the wrong answer when nobody in-house has time to ship anything, because a quarterly strategy session with no execution behind it produces a document rather than a citation. That is the one failure mode worth checking for before you choose it.

How much does AEO cost in-house compared with an agency?

The cost of AEO in-house is mostly salary and the cost of an agency is mostly fee, and every published figure for either comes from somebody with a position. Taking the three models in order of price: a tracker on its own starts at $29 a month for Otterly AI's Lite tier, which carries fifteen tracked prompts and four engines, with Claude, Gemini and Google AI Mode sold as add-ons on top, read from Otterly's own pricing page on 21 October 2026. Mid-tier agency-grade tracking runs from $95 a month and technical platforms from $250, and the dated figures for each are set out in our comparison of AEO tools in 2026.

Managed programmes sit an order of magnitude above that. AEO agencies charge roughly $3,000 to $15,000 a month, according to the 2026 AEO and GEO pricing guide from Digital Elevator, and that guide also notes that most established agencies publish no rate card at all, so the first number you hear is shaped by what the supplier thinks you will pay.

The in-house figure is the one buyers have least to go on, and the most detailed published breakdown comes from Discovered Labs. It prices a functional AEO team as three roles on US base salaries: an AEO strategist at $120,000 to $160,000, a technical SEO engineer with structured data expertise at $100,000 to $140,000, and a content specialist trained in entity-dense writing at $70,000 to $90,000. On top of base salary it adds benefits and payroll taxes at 25 to 30% and recruiter fees at 20 to 25% of first-year salary, and it allows four to nine months from the decision to a fully productive team, concluding that "the realistic first-year in-house cost for a functional AEO team sits between $420,000 and $450,000, with no guarantee of pipeline output within that year" (Discovered Labs, AEO agency versus in-house cost breakdown, read 21 October 2026).

Read that as a ceiling drawn by an interested party rather than as a budget, for three reasons visible on the page itself. Discovered Labs sells the alternative, and the same article recommends its own agency retainer at €6,995 a month. The arithmetic does not reconcile: per the same Discovered Labs breakdown, read 21 October 2026, the three base salaries total $290,000, the article separately gives $280,000 to $370,000 for the same three-person people cost, and the first-year total is stated as $420,000 to $450,000. And the comparison puts a dollar in-house cost beside a euro retainer without naming an exchange rate, which flatters whichever side the reader is not checking.

The deeper problem with the figure is that it prices a full function rather than the decision in front of you. Almost nobody starts a first year of AEO by hiring three people. The common in-house shape is a fraction of one existing marketer's week plus a tracker, which is why the number that decides this is hours rather than headcount: how many hours a week, whose, and for how many months before the quarter takes them back. A realistic market view of what each route costs, without our own figures in it, is in our guide to what AEO costs.

Why does measurement alone rarely move the number?

Measurement alone rarely moves the number because the number moves when pages change, and most of what an engine cites is a page you own and can edit. In the SIGNALS pharmaceutical process skid study, between 73% and 89% of citations pointed to firms' own pages rather than to directories or listicles, and the band held across six separate categories (SIGNALS, 2026). That puts the lever on your own domain rather than in third party placement, which is why the model you choose is really a decision about who is allowed to edit those pages and how often they get round to it.

So a tool that only measures, an in-house owner who only tracks, and an agency that only audits all stop in the same place. Each one produces an accurate description of a problem that nobody has fixed. When you compare the four models, the useful question of each is not what it reports but what it changes, and how many weeks pass between the report and the change going live. A model that shortens that gap is worth more than one that produces a better chart.

Frequently asked questions

Frequently asked questions about AEO cost and tooling

Related guides

• How to choose an AEO agency — the criteria that separate a real practice from a repackaged SEO shop.
• Best AEO tools in 2026 — the tools side of the comparison set, ranked and priced.
• How to track AI visibility — what good measurement actually looks like.
• The 7 SIGNALS dimensions and the research methodology behind them.