How to sell GEO services: the agency playbook

How do agencies sell GEO services? The five-step summary

Selling GEO services is a delivery problem, not a demand problem: 54% of businesses expect their existing SEO or marketing partner to lead the work, and 94% of CMOs plan to increase AEO/GEO budgets. The playbook that turns that demand into retainers has five steps:

1. Verify the SEO and technical foundation.
2. Deliver a white-label audit document that speaks to management and developers at once.
3. Replace keyword lists with business-relevant prompt sets.
4. Set up LLM monitoring — broad across engines for visibility, or deep on one engine for content strategy.
5. Report proactively with stakeholder-level dashboards and email sendouts.

Steps 1–2 win the client; steps 3–5 are the recurring revenue.

The shift: what selling search services means after AI search

For two decades, the agency business model was stable: rank the client for keywords, report positions and traffic, renew the retainer. AI search didn't destroy that model — it grew a second half that most agencies haven't productized yet.

Notice what didn't change: the client still buys visibility, and visibility still stands on a technical foundation. What changed is the buying trigger. Clients rarely walk in asking for "Generative Engine Optimization" — they walk in with one sentence of pain: "We asked ChatGPT about our category and it recommended our competitor." The agencies winning GEO retainers right now aren't the ones with the best acronym explanations. They're the ones with a rehearsed answer to that sentence.

There's also a positioning decision to make before the first pitch. You can sell GEO as an add-on to the existing SEO retainer — the easier upsell, because trust and access already exist — or you can reposition the agency around AI visibility entirely, which we've covered in From SEO agency to AI visibility agency in 90 days. Both paths run on the same delivery pipeline. That pipeline is the rest of this article.

The playbook at a glance

1. Verify the foundation — SEO basics and the technical audit come first, because everything else multiplies against them.
2. Deliver the audit as a white-label document — one deliverable that convinces the CMO and instructs the dev team.
3. Build the prompt strategy — replace the keyword list with business-relevant prompts that mirror how buyers actually ask.
4. Set up LLM monitoring — broad across the major engines, or deep on one engine, depending on what the client is buying.
5. Report proactively — stakeholder-level dashboards and email sendouts that make progress visible between wins.

Steps 1–2 are the door opener and the first invoice. Steps 3–5 are the retainer. One-off revenue funds your acquisition; recurring revenue is your business.

Step 1 — Start where SEO never stopped: the foundation

GEO doesn't replace SEO fundamentals — it raises the price of failing them. Generative engines run on the same substrate classic search does: a site that machines can crawl, render, extract and trust. The difference is tolerance. Googlebot renders JavaScript and forgives messy markup; most AI crawlers fetch raw HTML and forgive nothing.

That's why every GEO engagement starts with the technical audit, worked tech-first, blocker-first: anything that prevents a bot from accessing or processing content gets fixed before any content or prompt work begins, because an access blocker silently multiplies every other investment by zero. We've published the full pipeline — bot access, rendering, chunkability, trust signals — in the technical GEO checklist, so we won't repeat it here.

The client-facing point matters more than the technical one: never sell GEO work on top of a broken foundation. If the client's WAF is blocking GPTBot, three months of content optimization will produce nothing, the client will conclude GEO doesn't work, and you'll lose the retainer and the reference. The audit protects their spend and your credibility at the same time — which is exactly why it's the first thing you should get paid for.

Step 2 — Turn the audit into your first paid deliverable

An audit that lives in a crawler UI is homework. An audit that lands as a branded document is a sales asset. With bubbles1, the technical SEO/GEO audit — 80+ checks spanning both the classic and the GEO layer — generates as a professional, white-labeled audit document: your agency's logo, your colors, your name on the cover. The tool stays invisible; the expertise on display is yours.

The document is deliberately built for two audiences at once:

• The high-level management overview — scores by dimension, the three to five findings that matter, each tied to business impact, written for the person who signs. These are the pages that get screenshotted into board decks and forwarded to the CMO.
• The implementation detail for the dev team — affected URLs, exact specifications, prioritized fix lists. This is what turns the document from a diagnosis into a workplan, and it's what makes the client's developers respect your agency instead of resenting it.

That dual structure is the whole trick. A report only executives understand produces agreement but no action; a report only developers understand produces action but no budget. One document that serves both is what converts an audit into an engagement.

Pricing the audit follows the patterns we detailed in the technical GEO audit as a door opener: full consulting price for inbound leads, a discounted door-opener rate for outbound acquisition, or free — reserved for the accounts where the retainer justifies the giveaway.

Step 3 — Keywords vs prompts: build the prompt strategy

Here the GEO retainer properly begins, and here most agencies stumble — because they port their keyword thinking directly into the prompt era.

Keywords measure search demand: short, high-volume, decontextualized. Prompts capture conversational buying intent: long, specific, loaded with context. "Running shoes" is a keyword. "What are the best running shoes for flat feet under €150?" is a prompt — and it's the second one that decides whether an AI engine recommends your client.

A business-relevant prompt set is not a list of 500 generic questions run through every engine. It's a curated set mapped to the client's actual revenue lines:

• Category recommendation prompts — "best [category] for [audience]" — where inclusion equals consideration
• Comparison prompts — "[client] vs [competitor]", "alternatives to [competitor]" — where framing is won or lost
• Problem prompts — the questions the client's product answers, phrased the way real buyers phrase them
• Buying-guidance prompts — budget, use-case and constraint variations that mirror real purchase research

Generic prompt lists feel thorough and measure nothing — they burn monitoring budget on questions no buyer asks. (Why that happens, and how intent-mapping fixes the economics, deserves its own article; it's coming.)

This is the step PromptDNA was built for. At a high level: it generates business-relevant prompt sets from the client's context, maps them by intent so you can tell consideration prompts from comparison prompts, translates them for multi-market clients, manages them as reusable libraries you can adapt across clients in the same vertical, and runs them in configurable iteration runs — because a single response from an LLM is an anecdote, and repeated runs are data. Each of those capabilities gets its own deep-dive post later; for the playbook, what matters is that prompt strategy becomes a repeatable deliverable instead of a brainstorm.

Step 4 — Set up LLM monitoring: broad or deep?

Every client conversation about monitoring eventually reaches the same fork, and it pays to name it explicitly in your proposal. Clients want one of two things:

Option A — broad visibility tracking. Continuous monitoring across the major LLMs — ChatGPT, Perplexity, Gemini, Claude and more. The question it answers is managerial: where do we appear, how is it trending, how do we compare to competitors? This is the panorama view — the one brand teams and executives want on a dashboard.

Option B — deep single-LLM monitoring. Close tracking of one engine — usually ChatGPT, as the market-share leader — with tighter feedback loops, used to fine-tune the content strategy: which prompts we appear in, which sources get cited, what changed after each content push.

Here's the honest guidance most vendors won't give you: content strategy is not written per-LLM. The work that improves visibility in one engine — cleaner structure, clearer entities, citable self-contained passages, answer-first content — carries over to the others, because every engine is trying to solve the same retrieval problem. For strategy purposes, collecting data from a single LLM is usually enough. Broad tracking is a reporting and reassurance product; deep tracking is an optimization product. They're both legitimate — but they're different line items, and clients respect an agency that can explain the difference.

The practical default we see working: start deep on ChatGPT for the strategy work, then add broad tracking when management wants the panorama — or lead with broad tracking for enterprise brand teams and upsell the optimization loop later. In bubbles1, both modes are configured per client and per project in the AI Search Integration module, so switching or combining them is a settings decision, not a re-platforming.

Step 5 — Report proactively, or the retainer dies quietly

GEO results build gradually — citations accumulate, share of voice shifts over weeks, not days. That's a commercial risk: value that's delivered but never seen doesn't renew. The retainer killer isn't bad work; it's invisible work.

The answer is proactive, audience-shaped reporting:

• Shareable dashboards with stakeholder levels. The C-level sees share of voice, trend and competitive position. Marketing sees prompts, content actions and wins. The dev team sees the technical fix pipeline and what's still blocking. One data source, three views — because a report that speaks to everyone convinces no one.
• Email sendouts with the most relevant data. The dashboard nobody logs into loses to the report that arrives. Scheduled sendouts put the right numbers in the right inbox before the client thinks to ask — which is precisely the difference between an agency that reports and an agency that manages.

In bubbles1, this is the job of the intelligent dashboard organizer — the reporting app that ties the playbook together. It provides white-labeled, customizable templates that you fill by selecting the relevant metrics, charts and data directly from the dashboards and placing them on the client report. Define it once per customer — which stakeholders, which metrics, which cadence — and reporting runs itself. Your team's hours go into interpretation and commentary, where clients actually perceive value, not into copy-pasting charts into slides.

(What the monthly report should contain to renew retainers by itself is its own topic — that post is next in the series.)

Packaging and pricing the service

The market has already converged on a three-tier pattern, and it maps exactly onto the playbook:

• The door-opener audit (one-off): €990–2,500 discounted for outbound acquisition, €5,000–15,000 at full consulting price for inbound — steps 1–2.
• The GEO add-on to an existing SEO retainer: typically €500–2,000 per month per client — prompt strategy plus single-LLM monitoring and reporting, steps 3–5 in their lean configuration.
• The full AI-visibility retainer: €2,000–10,000 per month — broad multi-LLM tracking, the complete reporting setup, and content operations on top.

The margin logic is what makes this the healthiest service line an agency can add right now: data collection, auditing and report generation are automated, so senior time concentrates on interpretation, strategy and the client relationship — the parts clients can't get from a tool and won't get from a cheaper competitor.

One rule on free work: give away the snapshot, never the plan. A first visibility check on a prospect's brand costs you minutes and opens doors. The prioritized roadmap of what to do about it is the product. Keep them separate and the free work sells the paid work instead of replacing it.

FAQ

Q: Do we need to monitor every LLM for every client?
A: No. Content strategy is engine-agnostic — the structural and entity work that improves visibility in ChatGPT carries over to Perplexity, Gemini and Claude. Deep monitoring of one engine (usually ChatGPT) is enough to steer strategy; broad multi-LLM tracking is a reporting product for stakeholders who want the full panorama. Sell them as different line items.

Q: How much should we charge for GEO services?
A: Market benchmarks in 2026: one-off audits from €990 (door-opener) to €15,000 (full consulting), GEO add-ons to existing retainers at €500–2,000/month, and full AI-visibility retainers at €2,000–10,000/month depending on scope and client size. Position against the cost of invisibility, not against tool prices.

Q: Can we sell GEO without deep technical expertise in-house?
A: Yes — the data collection and first-pass prioritization are automated, and an expert audit review can be booked on top when a senior second opinion is needed. What the agency must own is strategy, interpretation and the client relationship. That's also where the margin is.

Q: How fast can we show a client results?
A: Technical fixes show effect within weeks — unblocked crawlers and repaired rendering are binary wins. Visibility and citation improvements typically take 4–12 weeks depending on content velocity. Put that timeline in the proposal; expectation-setting is part of the service.