GEO for agencies: what to actually tell a client who asks about AI visibility

A client emails: "hey, are we even showing up when people ask ChatGPT this stuff?" If you run an agency and you don't already have a real answer ready for that message, you're not alone — most of the industry doesn't yet, because until recently there was no equivalent of Search Console for it. This post is the answer worth having ready: what GEO actually is, a real check you can run live in the next call, and how to price the ongoing work so it isn't a favor you do once and never bill again.

The conversation you're probably already having, badly

Three answers tend to come out when a client asks this cold, and none of them hold up. "We're doing SEO, which helps with everything" is half true and doesn't actually answer the question — ranking and being recommended are mechanically different things, and a client who's heard that AI search is eating their category deserves a sharper reply than a shrug dressed up as reassurance. "Let me go check ChatGPT for you" is a real instinct but a weak measurement — one person typing one prompt once tells you almost nothing, because these models vary answer to answer and differ engine to engine. And silence — just not addressing it until the client brings it up again — is the most common answer of all, and the worst one, because it reads as not having thought about it rather than not having had the tool for it.

What GEO actually is, in one honest paragraph

Say this part plainly, because clients can tell when they're being sold a rebrand: Generative Engine Optimization isn't SEO with a new acronym. SEO competes to be a link on a results page; GEO competes to be a name in a generated answer, and that changes what "winning" even means — there's no page two, no fifth-place click, just named or not named. The evidence these engines lean on is different too: heavily weighted toward third-party proof — reviews, comparison pages, listicles, forum threads — rather than a brand's own homepage copy, which is the actual mechanism behind "AI recommends my competitor, not me" far more often than any quality gap. And the part that makes this a genuinely new discipline, not an old one renamed: when a model leaves a brand out of an answer, there's no referrer, no log line, nothing in the client's analytics that shows it happened. A client can be losing this quietly for months and never see it in a single existing dashboard they already check.

The 5-minute audit you can run live, in the actual meeting

This is the part that does more convincing than any slide: open a free PingMyBrand scan on the client's own domain, on screen, while they watch. No signup, about a minute, and it puts real buyer-shaped questions — not "what is [client]," but "what's the best [client's category] for [their actual customer]" — to ChatGPT, Claude, Gemini, and Perplexity, and reads back, question by question, whether the client was named. Then open the "named instead" view: if the client wasn't named, the slot almost never sits empty — a specific, real competitor is named there instead, and seeing that name out loud in the room is usually the moment the conversation shifts from "is this a real thing" to "what do we do about it." Read how the visibility score is actually built before this call so you can explain the number honestly if asked, rather than treating it as a green light or a grade.

Price it as a retainer line, not a report you hand over once

The mistake worth avoiding is treating this as a one-time deliverable — a PDF you generate, bill for, and never revisit. GEO needs the same ongoing posture rankings did once you realized they mattered: engines get retrained, competitors publish the comparison page you didn't, and a score that looked fine in March can slip by June with nothing in the client's existing tools to warn them. The Agency plan is built for exactly that ongoing shape: $199/mo covers 10 client brands, each re-scanned weekly across all four engines with 50 buyer questions per brand, white-labeled under your own agency name and logo — not ours — so the client sees your brand on the report, not a vendor's. That's $19.90/client/mo in tool cost before you bill a cent of it, and 3 read-only team seats so an account manager can pull a client's numbers without touching your login or billing. There's also a real, live referral mechanic worth knowing about: refer another agency or consultant and each redeemed referral adds a bonus client-brand slot on top of the base 10, at no extra cost — the per-client cost keeps dropping the more of your peers you bring in, not just in theory.

What not to do with this

Don't oversell it. Don't tell a client "AI recommends your competitor" without showing them the actual sentence an engine returned — a claim like that with no receipt attached is exactly the kind of thing that erodes trust the first time they check it themselves. And don't manufacture urgency with a number you haven't verified live on their own domain; the free scan takes about a minute, so there's no reason to guess when you can just show them the real one.

Where to send them, and where to start

/agencies has the full plan details, the white-label mechanics, and the FAQ a client-facing team usually asks before adding this as a line item. If you haven't run the audit above on your own agency's domain yet, start there — run the free scan on yourself first, no signup, about a minute, and you'll walk into the next client conversation with a real answer instead of a shrug.