Your Agency Says They Already Do GEO. Ask Them These Six Questions
By
Izzy Gregorio
·
5 minute read
Position · Screening
Your Agency Says They Already Do GEO. Ask Them These Six Questions.
Eighty-eight percent claim the capability. Thirty-seven percent of those admit the offering is not defined yet. One question sorts most of the market.
In short
Eighty-eight percent of organizations claim GEO capability and 37 percent admit their own offering is loosely defined, per GNW Consulting and Demand Metric's 2026 survey of 225 business-to-business leaders. The first question sorts most of the market: ask to see the prompt set. A provider doing this work has one. A provider repackaging SEO does not.
The state of it
Why screening is now your job
The GNW Consulting and Demand Metric study surveyed 225 business-to-business marketing and revenue leaders in 2026 and found four numbers that describe the market precisely.
| Finding | Figure |
|---|---|
| Organizations experimenting with or operationalizing GEO | 92% |
| Organizations claiming GEO capability | 88% |
| Of those, admitting the offering is loosely defined | 37% |
| Organizations with a dedicated GEO owner | Fewer than 15% |
Demand is real, capability claims are near-universal, and ownership barely exists. In a market shaped like that, credentials do not separate providers because everyone has the same ones.
What separates them is whether they can show you the instrument.
The questions
Six, in order of how much they reveal
- 1
Show me your prompt set.
Not a description of the methodology. The actual list of prompts, frozen, with dates. This single question sorts most of the market, because a provider measuring AI visibility has a file and a provider reselling a dashboard has a slide. If the answer involves explaining why they cannot share it, you have your answer.
- 2
Which engines do you test, and how do you weight them?
The correct answer names at least four and explains why. Engine behavior varies enormously. SOCi's 2026 Local Visibility Index found ChatGPT surfaces a local business 1.2 percent of the time against Gemini at 11 percent, so a single-engine program produces a single engine's result.
- 3
What is my baseline number, and when was it recorded?
If work has already started without a documented Day 0 figure on a frozen prompt set, nothing that follows is measurable. This question frequently ends the conversation, and that is useful information delivered early.
- 4
How much of your scope is off-site?
Roughly 82 percent of AI citations come from earned media, per Muck Rack's December 2025 analysis. A scope that is entirely on-site content production is aimed at the smaller half of the problem. Ask what percentage of hours goes to citation supply, listings, coverage, and original data.
- 5
What happens in month one?
The right answer is unglamorous: baseline test, crawler access verification, indexation check, entity and listing consistency, prioritized fix list. If month one is content production, the diagnosis was skipped.
- 6
What are you not able to promise?
The most revealing question in the set and the one nobody asks. A provider who has actually run these programs can tell you immediately: they cannot control third-party publishing timelines, they cannot guarantee specific engine behavior, and they cannot compress citation supply below a few months. A provider with no limitations to name has not done the work.
Translate it
What the answers mean
| If they say | What it usually indicates | What to do |
|---|---|---|
| We use an industry-leading platform, and cannot show the prompts | Reselling a monitoring subscription | Ask what they do with the data. The tool is not the service |
| We test ChatGPT | Single-engine program | Ask for the other three, and expect the price to reflect the work |
| We will establish a baseline once we start | Baseline treated as an afterthought | Insist it precedes any content work and is recorded in writing |
| GEO is included in your content retainer | Repackaged SEO with a new label | Ask which line items changed when GEO was added. If none, nothing did |
| We guarantee first-page AI visibility in 30 days | Marketing instrument | Ask which prompts it is measured against and who chose them |
| Here is our prompt set, here is your Day 0, here is what we cannot promise | They have done this before | Move to scope and pricing |
The pricing signal
What the number tells you before the proposal does
Published market ranges put mid-market GEO retainers at roughly $5,000 to $10,000 per month, boutique specialists at $5,000 to $15,000, and audits at $1,500 to $5,000. Those are market compilations rather than research, so treat them as orientation.
The useful part is not the range. It is what sits underneath a number far below it. Four engines, a maintained prompt set, structural editing, and any real citation supply work carry a cost floor, and the market openly describes offerings priced well under it as repackaged SEO with an AI label on the report.
A price far below the market range is a scope statement, whether or not anyone says so out loud.
One test that applies to any provider, including this one. Ask whether their pricing is stated the same way on every page of their own website. A provider whose numbers move depending on which page you land on has a scope problem before you have hired them.
The uncomfortable version
Your current agency may be fine
This post is not an argument for firing anyone, and a reader who has an existing relationship deserves the honest version.
Plenty of good agencies are early in building this capability and are transparent about it. An agency that says the technical and content work is strong, the off-site citation work is new, and here is what it can and cannot do yet, is more trustworthy than one claiming full capability across the discipline.
The failure mode is not inexperience. It is inexperience presented as expertise, and the six questions surface that distinction without requiring you to know the answers yourself.
One question worth sitting with. Send question one to your current provider this afternoon. Ask to see the prompt set.
Whatever comes back, you will know more about what you are buying than you did this morning, and it will have cost you one email.
Common questions
Evaluating a GEO provider, answered
How do I know if my agency actually does GEO?
Ask to see the prompt set. Not a methodology description, the actual frozen list with dates. Providers measuring AI visibility have a file. Providers reselling a dashboard have a slide. That single request sorts most of the market in one exchange.
What should a GEO scope include?
A documented baseline on a frozen prompt set, multi-engine testing, technical access verification, entity and listing consistency work, structural editing of existing pages, and off-site citation supply. A scope that is only on-site content production addresses the smaller half of the problem.
What GEO pricing is a red flag?
Published market compilations put mid-market retainers at roughly $5,000 to $10,000 monthly and audits at $1,500 to $5,000. Offerings priced far below the range are widely described as repackaged SEO with an AI label, because four engines, a maintained prompt set, and real citation supply work carry a cost floor. A price far below the range is a scope statement whether or not it is said out loud.
What reporting should I expect from a GEO agency?
Monthly, against a frozen prompt set, showing share of model, citation rate, and the competitive citation gap against named competitors. Quarterly is too slow: Amsive found half of all AI-cited content is under 13 weeks old, and AirOps found pages unrefreshed for three months are more than three times as likely to lose citations entirely.
Should I fire an agency that is new to GEO?
Not necessarily. Inexperience stated plainly is more workable than inexperience presented as full capability. An agency that names what it can and cannot do yet is often the better partner. The problem is claimed expertise that cannot produce a prompt set or a baseline.
What is the single best screening question?
Ask what they are not able to promise. A provider who has run these programs answers immediately: they cannot control third-party publishing timelines, guarantee specific engine behavior, or compress citation supply below a few months. A provider with no limitations to name has not done the work.
Ask us the same six
Start with question three
A test you hand other people should be one you can pass. The AI visibility audit from Conspicuouz Creative Group is our answer to question three: a documented Day 0 figure on a frozen prompt set, run across ChatGPT, Claude, Gemini, and Perplexity, recorded before anyone touches your site. You get the prompt set and the findings whether or not you ever work with us, which also answers question one.
Free. And ask us question six before you commit to anything.