How much does AEO cost?
AEO is sold on three pricing models, and which model you accept matters more than the figure attached to it: a monthly retainer, a fixed-scope project, or performance pricing tied to an outcome. This page does not quote market rates, because this site has not surveyed them and an invented range is worse than no range.
Asking what AEO costs is usually a proxy for two better questions: what am I paying for, and how will I know whether it worked. The pricing model answers the first, and it determines whether the second is even possible.
| Model | What you are buying | Where it goes wrong |
|---|---|---|
| Monthly retainer | Continuing work: entity maintenance, content, measurement runs | Pays for activity, so it can continue indefinitely without a result |
| Fixed-scope project | A defined deliverable: an entity audit, a schema build, a measurement baseline | Ends before the slow part, since corroboration takes months |
| Performance pricing | An outcome, usually citations or mentions | Prices something the seller does not control, which pushes them toward manufacturing it |
Why this page has no dollar figures
Every published range for this service that we have seen traces back to an agency quoting its own rate card. This site has not run a survey of the United States market, and inventing a range would be exactly the failure it criticises elsewhere: a number with no method behind it, repeated until it sounds like a fact. If you want a figure, collect three quotes for the same written scope; that is a survey of one, and it is one more than most published ranges have.
What actually drives the cost
Four things, roughly in order of weight. How ambiguous your entity is to begin with, since disambiguating a common name across a dozen properties is slow work. How much of your subject matter is already covered by stronger sources. Whether you have anything a third party would want to publish, because corroboration cannot be bought without becoming advertising. And how many queries, engines and countries you want measured, which is the one line that scales linearly and predictably.
The promise that should end the conversation
Guaranteed citation in a named engine. Nobody controls whether ChatGPT or a Google AI Overview names a brand; the systems are non-deterministic and vary by country and session, as the runs published here show. A guarantee can only be met by manufacturing the appearance of the outcome, and the cost of that lands on you when it is noticed.
What a defensible proposal contains
A written scope, a baseline measurement taken before any work starts, a fixed set of queries and engines that will not change mid-engagement, and a reporting format that shows unfavourable runs. Without a baseline, nobody can later tell whether anything moved. The measurement protocol is published here so you can ask for it by name.
How to compare two proposals
Put them side by side on four lines and the price stops being the interesting part. Does each include a baseline measurement taken before any work begins. Do the query list and the engine list appear in writing, and are they the same list in month six as in month one. Does the report show runs that returned nothing. And what specifically happens if, after six months, nothing has moved. A proposal that cannot answer the fourth question has not thought about being wrong.
The in-house comparison
The measurement half of this work is genuinely cheap to run yourself: a fixed query list, a weekly slot in somebody’s calendar, and a spreadsheet with the same columns every time. The expensive half is the entity and corroboration work, which needs someone who has done it before and is the part worth paying for. Buying the cheap half and doing the expensive half badly is the most common way this budget gets wasted.