← Blog

A Four-Figure GEO Audit. Let Us Work Out What the Client Is Actually Paying For

Mateusz Rzetecki

The conversation I have had a dozen times this year always runs the same way. An agency gets an enquiry: "we want to be visible in ChatGPT, what does that cost". The answer is that it starts with an audit, and the audit runs into the thousands. Sometimes well into them, once competitor analysis and a board-ready presentation are in scope.

The client goes quiet, says they will think about it, and never comes back.

Not because the money is not there. Because they were asked for a four-figure sum for a document whose value they have no way to judge, on a subject they know one thing about: that it is new. From where they sit, it is a purchase made blind.

This piece is about repositioning that conversation - and about where the margin in a GEO service actually sits:

  • what makes up a four-figure audit quote and how much of it is one-off work
  • why diagnosis should be cheap and repeatable rather than expensive and single-use
  • where the agency earns: digital PR, brand building, on-site fixes
  • how to package it as a retainer instead of a one-off
  • what not to promise if you want to keep the client past the first quarter

What an expensive audit is made of

Break a typical quote down into hours, because that is where everything becomes visible.

  • Question set and manual querying - an analyst types dozens of queries into several systems and records the answers. A dozen or more hours if it is done properly and the measurement is repeated.
  • Technical analysis - robots.txt, headers, rendering without JavaScript, sitemap, structured data. Another several to a dozen hours.
  • Competitors - the same again for five brands alongside. A multiplier.
  • Write-up and presentation - the most expensive part, because it is done by someone senior.

Hours times rate produces exactly those thousands. The quote is honest. The problem is not the rate, it is the construction of the offer.

Because what does the client get? A photograph of one afternoon. Model answers vary, competitors keep working, and in a quarter the document is out of date. Finding out whether anything improved means paying the same amount again - so nobody does, and the whole story ends with a single PDF in a folder.

From the agency's side it is worse still: the most expensive item in the offer is also the hardest to sell, and it generates no recurring revenue.

Separate diagnosis from execution

The whole shift comes down to one thing: stop selling diagnosis as a project.

Diagnosis has three properties that call for a product rather than billable hours. It is repeatable (same question set, same checks), comparable over time (it only means anything once you repeat it), and uncreative (there are not two ways to check whether an X-Robots-Tag header is blocking a crawler).

Execution is the exact opposite: it needs people, relationships, judgement and experience. That is the only part worth charging an hourly rate for.

This is where we come in - and deliberately only into that first layer. AISearchInsight.io is not an agency and does not do implementation. We do the measurement and the report: your brand's presence across ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews, the GEO audit of the site, a comparison with competitors, a list of recommendations. It is priced per report, not per hour, and it sits at a fraction of what a bespoke audit costs - the current price list for your market is on the pricing page.

None of this means your analytical work is worthless. It means you stop opening with it. Diagnosis stops being the barrier to entry and becomes the first step a client agrees to without deliberating - the step that puts the problem in front of them in black and white.

See what diagnosis costs for your client

The report covers brand presence across the five AI engines, the GEO audit of the site and a comparison with competitors. Agencies buy it at partner prices.

See the partner programme

Where the agency actually earns

Once the report is on the table, the conversation shifts from "is this worth it" to "what do we do about it". That is where the service begins.

Building presence in the places that get cited

Models assemble answers from other people's pages: comparison sites, directories, trade media, forums, video. A brand mention in one of those weighs more in this game than another post on the client's blog.

The work you sell:

  • expert commentary and quotes for trade media - not a press release about a new logo, but a concrete position with numbers in it
  • the client's own data as press material - a study, a breakdown, a statistic from their market; the most effective way to get a brand cited without paying for placement
  • presence in comparison sites and trade directories - completing profiles, specifications, pricing, category description
  • reviews - collected continuously rather than in bursts before the season
  • video - brand presence in YouTube material, including other people's: the transcript is content models read
  • a consistent category description everywhere the client has control: site, profiles, press materials

This is classic digital PR, with a different criterion for choosing the places. You no longer ask only about reach and link metrics, but about whether a given domain shows up in model answers in this category at all. The report hands you the list of those domains, so choosing where to work stops being a matter of instinct.

Fixes on the client's site

The second half of the work is technical and hands-on, so it belongs with the agency:

  • access for AI crawlers - three layers of blocking (robots.txt, the X-Robots-Tag header, meta robots), separating visibility crawlers from training ones, checking that bot protection is not cutting everything off
  • content visible without JavaScript - the most important pages have to deliver their substance straight from the server
  • structured data and the entity - Organization, Person for authors, sameAs to profiles, consistent naming
  • citable form - the answer up front, concrete claims, pricing tables, comparison blocks, an explicit "who this is for and who it is not for"
  • freshness - updating existing URLs instead of producing further posts alongside them
  • sitemap and basic performance - dull, quick, doable in one sprint

None of this is a revelation to a team that has done SEO for years. What is new is the order and the justification - and that is exactly what the report supplies: not "let us tidy things up", but "this particular header shuts out four crawlers, and once JavaScript is subtracted a quarter of this page's content remains".

How to package it

The model that works in practice has three parts:

  1. Entry: diagnosis plus a workshop. The report as a material cost, plus your few hours to walk through the findings and set priorities. The client gets something concrete at a price that does not require board approval.
  2. Retainer: execution. Digital PR, content, technical fixes, presence in directories and review sites. This is your time, your relationships and your margin. Price it like any other service.
  3. Periodic measurement: the re-check. A refreshed report with a before-and-after view, built into the retainer as part of reporting. That closes the loop: the client sees what changed, and you have something to show that the work mattered.

The third point matters more than it looks, because it solves the oldest problem with this service: how to demonstrate the effect of visibility work when there is no ranking position to paste into a spreadsheet. A before-and-after view on the same question set is an answer the client understands without translation.

What not to promise

This field is young and full of promises that come back around. Three that cannot be defended:

  • Positions. A model's answer has no position and no standing ranking. There is presence or absence, against a specific question, on a given day.
  • Timelines. Some changes show up at the next measurement, others need the world to write about the brand first. Committing to a calendar invites a difficult conversation later.
  • "Getting added to AI". There is no panel where you submit a brand to ChatGPT or Gemini. There is only what the models find.

Sell what you actually control instead: the scope of the work, its quality and regular measurement. That is enough to keep a client for longer than one PDF.

There is a separate track for agencies. We do not do implementation, so we are not competing for your clients - on the contrary, clients who come for a report and then look for someone to do the work get pointed to trusted partners. Partners get reports at partner prices, the re-check with a before-and-after comparison for reporting, and material for sales conversations. Joining costs nothing; you pay only for the reports you buy.

Buy reports at partner prices

No joining fee and no subscription. The report for your audits and pitches, the re-check for reporting results, and client referrals from our side.

Become a partner

The best part of this arrangement is that you stop defending the price of an audit. Diagnosis costs little enough that nobody deliberates over it, and from the first minute the conversation is about what needs doing - which is to say, about your work.

Mateusz Rzetecki

Brings over 16 years of experience in crafting and executing innovative SEO and content marketing strategies to enhance brand visibility and drive organic growth across diverse industries.