How to evaluate a GEO proposal
Score a GEO proposal on five things before you compare price: baseline, engines named, deliverables, third-party targets and how citations get measured.
Evaluate a GEO proposal by scoring five sections: whether it starts from a measured baseline, which engines it names by name, what it delivers month by month, which third-party sites it will pitch, and how it reports citations. A proposal missing two or more of these is not ready to compare on price yet.
This follows directly from how to choose a GEO agency: once proposals actually arrive, here is how to read them rather than just react to the number on the last page.
On this page
What's the fastest way to tell a proposal was actually written for you?
Look for your own website in it. A proposal built for your situation names pages it already looked at, cites a real technical issue it found on your site, or references a competitor by name. A proposal that could be sent to any company in your category with the name swapped is a template, and templates are not automatically bad — they just mean the real thinking has not started yet, and you should expect a second, more specific document before you sign anything.
If you ran an RFP first, this check is easier: what to put in a GEO RFP sets out the questions a response should be answering directly, and a proposal that ignores half of them is telling you something on its own.
Read the opening pages with a specific test in mind: could you swap your company's name for a close competitor's and have the document still make sense? If the answer is yes, the agency has not yet done the work of looking at your actual pages, your actual crawler status or your actual category, and everything specific still lies ahead of this document rather than inside it. That is a fair thing to say back to them directly before moving forward — a good agency will treat the question as reasonable, not as an insult.
How do you check the engines and baseline sections?
Look for names, not categories. ChatGPT, Claude, Gemini and Perplexity are reached through overlapping but different work — different crawlers, different citation behavior, different third-party sources they trust [1][2]. A proposal that says "AI platforms" without naming which ones has not actually scoped the work yet, whatever the price on the next page says.
The baseline matters just as much. Ask whether the agency ran, or plans to run, a first pass of your buyer questions across each named engine before proposing fixes. Without that starting number, there is nothing for month three's report to be measured against — you would be taking the agency's word for improvement rather than checking it.
Some proposals fold the baseline into the price of the engagement itself, running it in week one as the first deliverable; others run a short paid diagnostic first and use its findings to shape the proposal. Both are reasonable. What is not reasonable is a proposal with neither — one that jumps straight from a general pitch about AI search into a page-fix plan with no stated starting point, because that plan was written without knowing what actually needs fixing on your specific site.
What should the deliverables section actually commit to?
Specifics you could check against an invoice, month by month — not a paragraph about strategy.
A written baseline, delivered in week one or two
Current crawler status and a first read of your citation rate, not a promise to "assess" at some point during the engagement.
A page-fix count or list, not just "ongoing optimization"
How many pages, which ones first, and roughly what changes — pricing, comparison or specification pages usually go first, since assistants reach for those kinds of pages most.
A named number of third-party pitches per month
Even a rough figure like "eight to twelve outreach emails a month" is checkable later; "active outreach" is not.
A reporting cadence with a stated format
Monthly is standard. Ask to see a sample report attached to the proposal itself, not promised for later.
Read these four against each other, not just individually. A proposal can score well on any single one and still be weak overall — a detailed page-fix list paired with a vague, one-line outreach plan is a proposal that will show fast on-page movement and then plateau, because the slower-moving lever, third-party placement, was never really scoped.
How should you score pricing against scope?
Only after you have listed the deliverables side by side, not before. A number on its own tells you nothing about whether it is a good price, because it is not yet attached to a quantity of anything.
A proposal that only states a monthly fee, with the deliverables described in general language above it, is asking you to price the number before you know its shape. Go back to the deliverables list first and count what is actually being promised — pages, pitches, a report — before you form an opinion on whether the fee attached to it is reasonable for your budget.
| Proposal claim | What should back it up |
|---|---|
| "We'll optimize for AI search" | Named engines, and different tasks listed for each one |
| "Ongoing content improvements" | A page count or list, prioritized by which pages assistants already reach for |
| "We'll build AI visibility" | Named third-party sites or categories the agency will pitch, with a rough monthly volume |
| "Regular reporting" | A stated cadence, format, and a sample report attached |
| A single flat monthly fee | The fee broken down against the deliverables list above it |
What in a proposal should make you ask more questions before signing?
A specific citation number attached to a date. The same twenty to forty questions run twice in the same week can return meaningfully different citations from every major engine — one study tracking over thirteen thousand citations across four assistants found this churn in every engine it measured, not just the less mature ones [3]. A proposal promising "40 new citations by month three" is promising an outcome nobody controls, dressed up as a plan.
13,184 tracked citations
A fair proposal instead commits to a fixed prompt set, run on a schedule, with raw results shared rather than a summary. That is a process you can audit. A citation count by a date is a bet you are being asked to fund.
What if the proposal leans on case studies you can't verify?
Ask who the client was, and whether you can see the raw numbers rather than a summarized chart. A proposal is entitled to describe its general approach through past work, but a case study with no named client, no dated numbers and no way to check it is closer to a testimonial than to evidence. That does not automatically mean the agency is lying — plenty of client agreements genuinely forbid naming the account — but it does mean the claim should carry less weight in your evaluation than something you can actually verify, like the questions in this checklist that you can test directly against your own site today.
How do you compare two proposals side by side?
Deliverables first, then price, then the two softer signals: how specific the language is, and how the agency handled any question your RFP asked that was inconvenient to answer. If you're also weighing whether this needs to be a new vendor at all rather than an addition to search work you already pay for, adding GEO to an existing SEO retainer is worth reading before you finish comparing. And once you pick one, what GEO agency onboarding actually looks like tells you what the first 30 days should contain, so you can check the engagement against what the proposal promised.
Put the two proposals in one document, side by side, with a row for each of the five sections above. It takes twenty minutes and turns a comparison you would otherwise be doing from memory — remembering which proposal said what, days apart — into something you can actually look at. Most of the real difference between two proposals becomes obvious the moment they are next to each other rather than read on separate days.
A proposal that survives this scoring is not a guarantee the work will go well. It is evidence the agency did the unglamorous part — measuring, naming, specifying — before asking for a signature, which is a reasonable thing to require from anyone before you pay them monthly for a year.
Questions people also ask
What's the single most telling section of a GEO proposal?
The baseline section. A proposal that skips measuring where you stand today before pitching a plan has nothing to compare month three against later, whatever the rest of the document promises.
Should a proposal name specific third-party sites it will pitch?
Yes, or at least the categories — review platforms, comparison posts, directories in your industry. A proposal with no named targets is guessing at outreach the same way an unbriefed pitch guesses at strategy.
How do I compare two proposals that price things differently?
Line up the deliverables first, then the price. A proposal that costs less but includes fewer engines or no third-party outreach is not a cheaper version of the same thing; it is a smaller purchase.
What if the proposal promises a specific number of citations?
Question it directly. Nobody controls what a third-party model retrieves at answer time, so a specific citation count promised by a date is a claim worth asking the agency to walk back before you sign anything.
Sources
- Overview of Google crawlers and fetchers — Google Search Central (checked 23 September 2026)
- Overview of OpenAI crawlers — OpenAI (checked 23 September 2026)
- I tracked 13,184 AI citations across 4 LLMs: here's what I learned — Orbit Media Studios (checked 23 September 2026)
Want a second read on a GEO proposal before you sign it?
Book a free 30-minute consultation with Ibrahem and Jeremy. We look at your site while we talk, and you leave with a prioritised plan.
Who wrote this
Jeremy ZavaCo-founder
Leads growth and client partnerships.
Works on: CRO, SEO
Reviewed before publishing by Ibrahem Helal, Co-founder.
- How to choose a GEO agency in 2026A buyer's guide to hiring a GEO agency: what the work is, what it should cost, how AI citations are measured, and the questions that separate two proposals.
- What GEO agency onboarding actually looks likeThe first 30 days with a GEO agency, phase by phase: the baseline audit, the access you'll need to grant, and what your first report should contain.
- How to write a GEO RFPA GEO RFP template: what to ask vendors about engines, current baseline, deliverables, third-party outreach, pricing and reporting before you compare proposals.
- Adding GEO to an SEO retainerWhether to ask your current SEO agency to add GEO work or hire a specialist, what to check they can do, and how the scope should be priced.