AI Visibility for agencies

What an AEO retainer sells

An AEO retainer sells one outcome: your client gets cited and recommended when buyers ask ChatGPT, Claude, Gemini, Perplexity, or Grok about their category. You deliver a measured baseline across those engines, a prioritized list of what is losing citations and why, the source content and structured data to fix it, and a monthly re-measure that shows the score moving. That is the whole offer.

Position it as adjacent to SEO, not a replacement. Most prospects already pay someone for search. Framing AEO as a second discipline with its own rubric, its own deliverables, and its own KPIs sidesteps the "so my SEO is wrong?" argument that kills the deal in the first meeting.

The concrete thing you are optimizing is measurable. Pondral's published rubric weights five factors: Presence (20), Prominence (25), Context (20), Citation Link (20), and Competitive Presence (15). Anchor your offer to a scorecard the client can read, not to vague "AI optimization." When the retainer has a number attached, renewal conversations get easier.

Lead with the audit, not the deck

The fastest-closing pitch is a baseline audit on the prospect's own brand, run before the first call. All five engines, ten to fifteen buyer queries in their category, real findings about where they get cited and where a competitor takes the answer. Nothing in a slide deck competes with a screenshot of ChatGPT recommending their competitor by name.

Keep the free audit cheap to produce so it scales across your pipeline. A tight scope is enough to earn the meeting:

1. Pick 10-15 queries a real buyer would type: category terms, "best X for Y," "X vs competitor," and one or two problem-first questions.
2. Run each query across all five engines, ChatGPT, Claude, Gemini, Perplexity and Grok, so the finding is not a single-engine fluke.
3. Record whether the brand appears, where in the answer it lands, and which sources the engine cited.
4. Screenshot two or three answers where a competitor wins outright. Those images are the emotional hook.
5. Summarize on one page: current standing, the biggest gap, and the first thing you would fix.

Scoping the engagement

Scope per brand, not per deliverable. A brand is the unit a client understands and the unit that scales cleanly across your book. Inside each brand, decide how many engines you monitor, how many queries define the query set, how often you re-measure, and how much source content you produce per month. Those four dials are your scope.

Write scope as an outcome plus a fixed monthly production capacity, so "more work" is a defined upgrade and not an open tab. A workable structure:

• Query set size: 15-30 queries for a single-brand engagement, more for multi-product or multi-market clients.
• Engine coverage: all five major answer engines by default. Add region-specific engines where the client's buyers actually are.
• Re-measure cadence: monthly for active retainers, quarterly for maintenance.
• Content output: a fixed number of source pages, comparison pages, or schema fixes per month, named in the SOW.
• Review scope: a standing quarterly session to re-cut the query set as the client's category and competitors shift.

Pricing models that hold up

Three pricing shapes dominate services work, and AEO fits all three. Retainer pricing is the default for anything ongoing: a fixed monthly fee for a defined scope, which gives you predictable revenue and gives the client a predictable line item. Project pricing suits a one-time audit or a fixed content build with a clear finish. Performance-linked pricing ties part of the fee to a movement in the score or in cited share-of-voice, which sounds appealing but only works once you have a clean baseline both sides trust.

As a general market reference, professional-services AEO and content retainers commonly run in the low-single-digit to five-figure monthly range per brand, scaling with query-set size, engine coverage, and content volume. Where you land depends on your market, your seniority, and whether you are producing content or only measuring. Price the scope, not the hours. Hourly billing punishes you for getting faster.

A few pricing rules that keep the P&L honest:

• Separate one-time setup (baseline audit, query-set design, schema fixes) from the ongoing retainer, so the first invoice is not fighting the recurring one.
• Tier by scope, not by feature access: a bigger tier buys more engines, more queries, and more content, not a longer feature list.
• If you offer performance pricing, define the metric, the measurement method, and the baseline in writing before the clock starts.
• Build your own tool cost into the floor price. If you are paying for a measurement platform per brand, that cost sits under every tier.
• Re-price at renewal, not mid-term. Lock scope for the term and adjust when the term rolls.

Reporting cadence

Report monthly, and lead every report with the score trajectory. Clients do not renew because you shipped four blog posts. They renew because their grade moved from a C+ to a B+ and you can show the line going up. Open with the chart, then explain what drove it, then say what you are doing next month. Activity logs belong in an appendix, if anywhere.

Standardize on a small set of numbers so the story stays consistent month over month. Three metrics carry most reports: the overall visibility score, cited share-of-voice against the tracked competitors, and gap-closure rate against the prioritized list. Those map to how visible the brand is, how it stacks up, and how fast you are executing.

Set the cadence in the SOW so it is a commitment, not a favor. A monthly written report, a live walkthrough each month or quarter, and a standing quarterly review to re-cut the query set is a defensible rhythm that most clients respect and few outgrow.

White-label deliverables

Your client should see your agency's name on the report, not your vendor's. White-labeled exports keep the relationship yours and keep the client from wondering whether they could just buy the tool directly. The report header, the color system, and the cover should read as your brand. The underlying measurement can come from whatever platform you run.

The deliverable set that reads as a real practice, not a spreadsheet dump:

• A branded monthly scorecard: overall score, per-engine breakdown, and the trend line since baseline.
• A prioritized gap list: what is losing citations, why, and the specific fix, ranked by expected impact.
• The produced content and schema itself: the source pages, comparison pages, and structured-data markup you shipped that month.
• A competitor view: where tracked competitors are winning answers the client should own.
• A short written narrative in plain language, so the client's exec can forward the report without translating it.

Show progress with reproducible evidence

The hard part of AEO reporting is that engine answers are non-deterministic. Ask the same question twice and the wording, and sometimes the recommendation, shifts. If your reporting treats a single run as gospel, one bad draw makes it look like you went backwards, and you will spend the QBR defending noise instead of results. Reproducibility is what makes the number defensible.

Build evidence the client could re-run themselves. That means fixing the query set and the engine list run to run, sampling each query more than once and reporting the average rather than a single answer, keeping the raw responses and the dates, and disclosing which engines were covered. When a score moves, you want to point at method, not vibes.

A practical evidence standard for agency reporting:

1. Freeze the query set and engine roster for the term. Note any change and the date it took effect.
2. Sample each query multiple times per run and report the average, so one odd answer does not swing the headline number.
3. Keep raw engine responses with timestamps, so any figure in the report traces back to a source you can show.
4. State the engines covered and the run date on the report itself. A line like "measured across five engines on July 3" beats an unsourced number.
5. Show the same queries month over month, so the trend is like-for-like and the client can watch a specific answer improve.

Handling the questions clients actually ask

Two questions come up in almost every scoping call. "Can you guarantee I'll rank #1 in ChatGPT?" No, and any agency that says yes is selling you a story. Engine answers change, and no one controls the model. What you can commit to is a measured baseline, a defined workstream, and honest monthly reporting on the trend. Sell the process and the evidence, not a placement you cannot control.

"How is this different from my SEO?" The overlap is real but the surface is different. SEO optimizes for a ranked list of links. AEO optimizes for whether a model cites and recommends the brand inside a synthesized answer. The tactics that help, clear source content, consistent entity data, structured markup, do overlap with good SEO, which is why the two retainers sit well side by side rather than competing. Say that plainly and the deal stops being a turf fight.