The question is already happening.
Last quarter, an agency I work with got a Slack message from a client at 9:47 on a Tuesday morning. A screenshot of a Perplexity result. The query was a generic “best [category] in [city]” question. Three sources cited. None of them was the client. Two were direct competitors.
The message ended with: “What are we doing about this?”
If you run an agency long enough, you’ve already had this conversation, or some version of it. AI Overviews started surfacing where blue-link results used to be. ChatGPT got a search button. Perplexity moved from curiosity to something a third of your clients now use weekly. And the question on every retainer call eventually arrives:
Why aren’t we showing up in AI answers?
This post is about how to handle that conversation. Not the diagnostic work itself — that comes after — but the conversation. Because the way an agency handles the first ten minutes of this discussion sets the entire scope of the work that follows.
What not to promise.
The fastest way to lose this conversation is to promise outcomes you don’t control.
Four overclaims I’ve seen agencies make in the last year, and why each backfires:
“We’ll get you cited in ChatGPT.” No agency can. Citation behavior is determined by the AI provider’s retrieval and synthesis logic, which is opaque, model-dependent, and changes without notice. What you can promise is to make the page available, structured, and trusted enough that being cited becomes possible. That’s the honest commitment. Anything else gets you fired the first time the client doesn’t appear.
“We’ll fix your AEO score.” Score-based commitments treat the diagnostic as the deliverable. Scores measure readiness, not citation. A client whose score climbs from 64 to 87 but who still doesn’t see themselves in AI Overviews will read that as a failure even when it isn’t. The commitment should be to specific structural and content changes, not to a number.
“We’ll get you in AI Overviews in 90 days.” AI Overviews are query- and context-dependent. The same site might appear for one query and not another, in one user’s context and not another’s. Time-bound citation guarantees collapse on first contact with reality.
“AEO will replace SEO for your business.” It doesn’t. SEO is the foundation. AEO sits on top. Pitching AEO as a replacement creates the wrong frame for everything that comes after.
The pattern across all four: mistaking what’s measurable for what’s controllable. You can measure citation. You can’t directly cause it. Treating those as the same thing is the structural error.
A four-question framework for scoping the conversation.
Before quoting any AEO scope, walk the client through these four questions. They take about fifteen minutes. They convert a vague “are we doing AI?” into a structured engagement.
1. Where is the client currently visible?
Not where they want to be — where they actually are. Branded search rankings, generic-query rankings, direct traffic, referral mix, current AI surface presence (if any). Establishes the existing visibility baseline. A client with strong branded search and weak generic-query rankings has different AEO leverage than one with the inverse.
2. Where do they expect to be visible in AI answers?
This is the question that separates real engagements from theatrical ones. Which queries? “Best [our category]” is a different problem than “how does [client] compare to [competitor]” or “what is [topic].” Which AI systems? AI Overviews and Perplexity weight differently than ChatGPT or Claude. Which user contexts? Logged-in vs. anonymous, mobile vs. desktop, geographic.
If a client says “we want to show up in AI” and can’t narrow further, you don’t have a scope yet. You have a brief.
3. What does their existing content infrastructure already make available?
Not what’s missing — what’s already there. Schema coverage, entity definition, author signals, page structure, robots.txt allowlists, content depth on the queries they care about. About 30–40% of agencies’ AEO work in the field is just surfacing what’s already on the site in machine-readable form. The other 60–70% is genuinely new content or structural work.
4. What’s the gap, and which gaps fit inside the existing retainer?
The gap is the delta between #2 and #3. The honest scoping move is to separate that gap into three buckets:
- Inside existing retainer scope (technical SEO with AEO-aware additions, content updates, schema work)
- Adjacent to existing scope (new content production, schema architecture, author profile work)
- Net new (cross-system monitoring, structured-data execution at scale, agency white-label reporting)
Quote each bucket separately. Don’t blend them into a single line item. The client’s response to the bucketing is itself diagnostic — agencies that get pushback on the “net new” tier are talking to clients who aren’t ready to expand the engagement, and that’s useful to know early.
How to scope the work.
Three scope tiers that fit inside most existing retainers:
Diagnostic. A one-time scan plus written readout. Useful when the client is asking the question but isn’t ready to commit to ongoing work. Positions the agency as the trusted reader of the diagnostic. Doesn’t require the agency to take on execution risk. Sets up the next conversation.
Remediation. Diagnostic plus a prioritized fix list plus execution. The bulk of agency AEO engagements land here. The honest framing is “we’ll close the gaps that are inside our craft to close, and surface the ones that need engineering or content work outside our scope.” Be specific about what’s inside the agency’s execution capability and what isn’t. Schema work, content rewrites, robots.txt changes, on-page structural fixes — typically inside scope. Headless-CMS backend changes, app-router rewrites, custom integrations — typically outside.
Ongoing. Monthly scanning, monthly fix execution, quarterly reporting on cross-system visibility. This is the retainer expansion pattern. Doesn’t replace the SEO retainer; sits beside it. The honest pitch: “the AI surface is a moving target, and a one-time fix list goes stale. Ongoing monitoring keeps the work current and gives the client a defensible track record of being prepared.”
The mistake most agencies make is collapsing these three tiers into one. The client doesn’t have a single AEO problem. They have a diagnostic problem, an execution problem, and an ongoing-monitoring problem, and not all of them need to be solved at the same time.
What to commit to.
The honest commitments an agency can make:
- Improving Access, Understanding, and Extractability scores across a defined set of pages, with a measurable baseline and a dated target.
- Closing specific schema and structural gaps identified in the diagnostic, with each gap tracked individually.
- Tracking where the site appears across AI systems over time, so the client has a defensible record of progress even when individual citations are noisy.
- Coordinating AEO work with existing SEO and content scope so nothing breaks and the foundational work compounds.
What an agency cannot commit to, even with the right diagnostic in hand:
- A specific count of citations across AI systems.
- A specific placement in any AI Overview, Perplexity result, or ChatGPT response.
- A specific traffic outcome from AI surfaces.
- A timeline for any of the above.
The credible commitment is to improve readiness.
The non-credible one is to guarantee outcomes.
This distinction is the entire conversation. Agencies that hold this line will keep clients longer than the ones that don’t. The clients can tell.
Close: a free scan as a discovery tool.
The natural opening move for any of these conversations is a scan. It runs in 90 seconds, it’s free, and it gives both sides a shared diagnostic to react to instead of debating in the abstract. The conversation moves from “is this real?” to “here’s what we found, here’s the gap, here’s the path.”
For agencies who want the broader context behind why this conversation is showing up across the industry, the founder note on why we built AIVZ covers the thesis. For SEO leads inside agencies who want to understand the technical bridge, the AEO-for-SEO-pros post is the right starting point.
If you’re heading into one of these conversations this week, run a scan on the client’s site first. You’ll be 80% ready before the call starts.