What a GEO programme costs, and what should be in the scope
Generative engine optimisation is being sold at wildly different prices for wildly different work. Here is what the deliverables actually are and what each tier should include.
Key takeaways
- Expect $1,500–4,000 for an audit and $1,500–5,000 monthly for an ongoing programme.
- An audit must include per-agent access testing, a no-JavaScript render check, an entity audit and a frozen baseline sample.
- A programme consisting mainly of new articles is a content retainer, not generative optimisation.
- If the site is not indexed or ranking, GEO is premature: search visibility is what feeds retrieval.
A generative engine optimisation programme typically costs between 1,500 and 5,000 US dollars a month, depending on how much content is being rewritten and how many languages are covered. A one-off readiness audit, covering crawler access, rendering, entity consistency and a baseline visibility sample, generally runs 1,500 to 4,000. Anything priced far below that is usually a formatting pass sold under a new name.
- $1,500–4,000
- one-off AI visibility audit with baseline measurement
- $1,500–5,000
- per month for an ongoing programme, depending on scope and languages
- 4–8 weeks
- typical time before assistant answers begin to shift measurably
What belongs in an audit
- Crawler access verification, including a live robots.txt read, per-agent request testing, and a check of CDN bot protection which blocks at the edge invisibly.
- Rendering check. What the page contains when fetched without JavaScript, since most AI crawlers do not execute it.
- Entity audit. Every place your organisation is described, compared for consistency, with the contradictions listed.
- Baseline visibility sample. A frozen set of about twenty commercial questions run across the relevant assistants, recording named, cited, position and who appeared instead.
- A prioritised remediation list with owners, not a scored report.
What belongs in a monthly programme
Ongoing work should be mostly editorial rather than technical, because the technical faults are finite and get fixed. Expect content rewrites that put direct answers first and add sourced figures, creation of quotable assets such as original data or clear definitions, entity corrections across third-party sources, and the monthly measurement sample with a written interpretation of what moved and why.
A programme that consists mainly of publishing new articles is a content retainer, which may be what you want but is a different product. GEO's distinctive value is usually in making existing pages citable, because those pages already have the authority that gets them retrieved.
Questions that separate real proposals from repackaged ones
Ask what specifically will be measured and how, since a proposal with no measurement plan cannot demonstrate value and will be judged on vibes. Ask what happens if the technical audit finds nothing wrong, because a well-built site often passes and the honest answer is that the budget moves to editorial. Ask which findings depend on current model behaviour and might change. And ask for the reasoning behind any recommendation that is purely about formatting, given that controlled research found formatting changes alone had weak effects.
When to not buy it
If your site is not indexed or does not rank for anything, generative optimisation is premature. Retrieval draws heavily on search visibility, so the sequence is search first, then generative. Spending on GEO before the fundamentals are in place buys you a better version of invisible.
GEO's value is mostly in making existing pages citable, not in publishing new ones.
Sources