AEO in-house vs agency

Key takeaways

• The knowledge is not the moat. Most of what an AEO agency does is public and has been written about at length.
• The failure mode of in-house AEO is not doing it badly. It is doing it for six weeks and then stopping.
• Cost comparison is against a salary, not against a tool. Software runs $100 to $200 a month; a person who will actually do the work costs far more.
• If you have someone senior with real capacity, in-house usually wins on cost and on context.
• If the honest answer is "we will fit it around everything else", that is the case for outsourcing, and it is a legitimate one.

Start by naming the constraint

There are three different problems that all present as "we are not showing up in AI answers", and they have different answers.

The first is a knowledge problem: nobody on the team knows what moves an AI citation. This is the easiest to solve and the least common. The material is public. A capable marketer can read for a weekend and know most of what matters.

The second is a capacity problem: someone knows what to do and cannot get to it. This is by far the most common, and it is the one people misdiagnose as a knowledge problem, because "we need an expert" is easier to say than "we deprioritised this four times".

The third is an ownership problem: several people touch it, nobody holds it, and the work exists as a recurring agenda item rather than as anyone’s job. This is the most expensive one, because it can persist for a year while looking like progress.

The honest case for doing it in-house

In-house genuinely wins in several situations, and an agency that will not tell you this is selling rather than advising.

• You have a senior person with real capacity. Not someone who will add it to a full plate. Someone with room.
• Your category is narrow and you know it cold. Context beats technique more often than agencies admit. Someone who knows the buyer can write a better comparison page than a specialist who does not.
• You are pre-product-market-fit. If the positioning is still moving monthly, engineering citations around it is premature. Fix the story first.
• Your problem is genuinely technical and you have engineers. If the blocker is that your site renders client-side and crawlers see an empty page, that is a sprint, not a retainer.

The honest case for an agency

The case is not that the work is arcane. It is that the work is continuous, unglamorous, and easy to drop.

AEO has an awkward shape for internal teams. It is not a project with an end date, and it produces very little visible output in the first six weeks. That combination makes it the first thing sacrificed when a launch slips, which means the sixth week is where most internal efforts quietly stop, right before the point where anything would have shown up.

An agency solves this by making the work someone else’s only job. That is most of the value. The rest is pattern recognition across accounts: knowing which surfaces matter for your category, which outreach gets answered, and which technical problems are worth fixing versus tolerating.

What it actually costs both ways

Compare against a salary, not against software. The tooling is cheap and getting cheaper: tracking products run roughly $100 to $200 a month, and you can approximate a lot of it manually. Owning a dashboard is not a competitive advantage for anyone, including agencies.

The real in-house cost is a person’s time. A rough test: AEO done seriously is somewhere between a third and a half of one competent marketer’s week, sustained. Price that fraction of a salary, add the tooling, and compare it to a retainer. Sometimes in-house is obviously cheaper. Sometimes the fraction is one you cannot actually free up, which is its own answer.

Ante publishes its rate at $6,000 a month partly so this comparison can be made before a sales call rather than after one.

A test that settles it in ten minutes

Answer these honestly. They are more diagnostic than any pitch.

1. Name the owner. Not the team. One person. If you cannot, that is the finding.
2. Ask what got dropped last quarter. If it was the slow compounding work, this will be dropped too.
3. Check the last attempt. Most teams have tried already. What stopped it? Whatever that was is still there.
4. Decide what you would cut to make room. If the answer is nothing, you have not made room.
5. Set the kill criterion now. What would make you stop in 90 days? Write it down before you start, in-house or not.

The hybrid most people should consider

The framing is rarely all or nothing. The parts that benefit most from an outside owner are the ones that are continuous and externally facing: third-party placement, outreach, weekly measurement, and keeping the technical floor from rotting. The parts that benefit most from staying inside are the ones that need product context: positioning, what is actually true about the roadmap, and which customer problems are worth writing about.

A reasonable split is that the outside party owns the surfaces and the cadence, and your team owns the substance. That fails when it becomes a coordination job for you, which is the thing you were trying to avoid, so it only works if the outside owner drives it.

The summary

Most companies could do this in-house. Most companies will not, not because they lack the skill but because the work has no natural owner and produces nothing visible for a quarter. That is a real constraint, not a failure of will, and it is worth taking seriously when you decide.

If you want to know where you stand before deciding anything, the free AEO report covers the same ground an agency would cover in a first month, and it does not commit you to a retainer.

Frequently asked questions

Can we just do AEO in-house?

Usually yes, on capability. The material is public and a competent marketer can learn it quickly. The question is whether one person can own it for two quarters without it being the first thing dropped when priorities shift. That is what determines the outcome, not skill.

How much time does AEO take in-house?

Done seriously, somewhere between a third and a half of one competent marketer’s week, sustained rather than in bursts. The sustained part matters more than the total, because the work compounds and stopping resets much of it.

Is an AEO agency worth it if we already have an SEO agency?

Sometimes, but ask your current agency first. Many strong SEO firms have added competent AI search work. The question is whether they treat it as the product or as a section in the monthly report, and whether they can show you what changed in an actual AI answer rather than in a rankings chart.

What is the cheapest way to start?

Ask ten buyer questions in ChatGPT, Claude, Perplexity and Google AI Mode, and write down who gets named and what is said about them. That costs an afternoon and tells you whether you have an absence problem, an accuracy problem, or neither.

How do we know if an agency is actually doing the work?

Ask for the citation itself, not the dashboard. A real report shows the question asked, the engine, the answer text, and where your name did or did not appear. Anything that only shows a score is measuring rather than delivering.