If a client has already asked you whether they show up in ChatGPT, you are not choosing whether to offer AEO anymore. You are choosing whether to sell it as a real, priced program or keep improvising it one favor at a time. This guide is about how to sell AEO services the right way: turning AI-search work into named packages with clear deliverables, a defensible price, and a pitch that gets a signature instead of a shrug. By the end you will have a package ladder, a pricing model, and a script for both existing and new clients.
This piece focuses on packaging, pricing, and the pitch. It does not cover how you actually deliver the work once it is sold, how to run the audit you might use as a lead magnet, or how to resell someone else's platform under your own brand. Those are separate questions with their own answers.
Step 1: Define the outcome you are actually selling
Before you name a single package, decide what you are charging for. Most agencies default to "AI content" or "we'll optimize you for AI," and both of those are too vague to price or defend. The market uses AEO, GEO, and AI-search optimization almost interchangeably, and you do not need to win the terminology argument with your client. Pick the term they already use, define it once in the proposal, and describe the real work underneath it. Agencies that manage to productize GEO services well treat this step as the foundation, not a formality, because everything you price later traces back to how clearly you defined the outcome here.
A workable definition: AEO is the repeatable process of improving how often a brand gets mentioned, cited, and represented in AI-generated answers to the questions its buyers ask. That definition breaks into four separate things you can actually sell, and separating them is what makes a package feel like a program instead of a grab bag.
- Visibility intelligence. Finding the buyer questions that matter, then tracking how often the client and their competitors get mentioned or cited in the answers.
- Content and technical work. Restructuring existing pages and writing new ones with answer-first structure, clear FAQs, schema, and entity clarity.
- Authority and citation building. The third-party sources, reviews, and expert placements that AI systems pull from when they answer.
- Measurement and iteration. Reviewing what changed, which prompts were won or lost, and updating the plan from that evidence.
How to tell you're done: you can say, in one sentence each, what your agency measures, what it builds, what it earns from other sources, and how often it reviews the results. Where agencies go wrong here is selling a bigger blog subscription and calling it AEO. A tracked prompt, a competitor citation, or a page-level win is the real unit of value, not another article on the pile.
Step 2: Choose the client and prompt scope before you scope the package
A package cannot be priced against "AI search" in general. It has to be priced against a defined slice of it: which markets, which products, which buyer stages, which competitors, and which AI platforms you are actually going to track. This is the scoping conversation that happens before you write a single deliverable line, and skipping it is why so many AEO proposals end up either overpriced for what they cover or underpriced for the work involved.
Ask the client (or figure out yourself, for an existing account):
- Which products, markets, and buyer stages matter most right now?
- What questions do prospects ask before they ever talk to sales?
- Which competitors show up most often in those questions?
- Does the client have pages worth improving, or does most of this need to be written from nothing?
- Who signs off on product claims and any regulated language?
- Who owns the CMS and the actual publishing?
You'll know this step is done when you can write a single sentence naming the markets, the rough prompt count, and the platforms in scope. That sentence becomes the spine of the package you build next.
Step 3: Build a ladder of three or four packages
This is where you package AEO services agency-wide into something a buyer can actually choose between, rather than a single quote that either takes it or leaves it. A ladder gives you room to serve a small account and a large one without inventing a new scope from scratch every time, and it gives the client an obvious upgrade path once the first package proves itself.
Foundation (or Visibility) package. For a first AI-search program, a smaller brand, or an account with no baseline data yet. It should include a defined set of core brand, category, comparison, and buyer-intent prompts, a baseline mention rate, citation rate, and share of voice, foundational optimization of a set number of existing pages, a small batch of new answer-oriented content, and a monthly report and review call. It should not include unlimited prompts, unlimited revisions, or every AI platform. Done means the client has an agreed prompt set, a baseline, a named competitor list, and a recurring report.
Growth (or Citation) package. For a mid-market brand that already has content but is losing the questions that matter to a competitor. Broaden the prompt mapping, add multi-platform monitoring, competitor citation analysis, a defined monthly volume of new articles and page refreshes, and a backlog of opportunities tied directly to the visibility data. Done means every recommendation in the monthly deliverable traces back to a specific prompt, competitor page, or citation gap, not a general sense that "more content helps."
Authority (or Market-Leader) package. For competitive categories, established brands, or multi-product companies that need authority work alongside content. This tier adds multi-market or multi-product prompt mapping, continuous competitive benchmarking, a larger content cadence, and authority work such as digital PR or expert placements where it's legitimate and the agency has the capability. Executive-level reporting belongs here too, along with a named process for who approves claims and handles anything reputationally sensitive.
Enterprise (or Custom) package. Scope this one with variables, not adjectives: number of brands, markets, and prompt groups; number of platforms monitored; content units per month; the digital PR and authority scope; approval and legal-review requirements; and what integrates with the client's own CMS or analytics. A package earns the enterprise label because of its complexity and risk, not because it costs more.
Whatever tier you're proposing, spell out ten things explicitly so nobody has to guess what "AI optimization" means: the platforms covered, the prompt scope, what counts as a content unit, whether authority work is in or out, the reporting cadence and metrics, what the client is responsible for, how many revision rounds are included, whether it's a one-time project or a retainer, what's excluded, and when you'll do a bigger strategic review versus a routine check-in.
Pro tip: treat the package table as a boundary document, not a marketing page. Every line either says what's in or what's out; nothing gets left implicit for a client to assume their way into. This is also the moment to decide, once, how your agency will package AEO services agency-wide for every future client rather than rebuilding the scope conversation from zero each time a new logo shows up.
Once the visibility and content layer is defined per tier, standardizing it across every client is the operational problem, not the strategic one. This is where a platform like DeepSmith earns its place in the stack: its AI Visibility module tracks mention rate, citation rate, and share of voice across the AI engines you're covering, with prompt history, a competitor leaderboard, and the pages actually getting cited, so you're not rebuilding a tracking spreadsheet and a report template for every new logo.
Step 4: Choose the pricing model on purpose
Once the package is scoped, pick how you charge for it deliberately instead of falling back on whatever the last client happened to negotiate. Each model fits a different kind of engagement, and mixing them without saying so is where margin quietly leaks.
Monthly retainer. The natural default for ongoing AEO, because visibility monitoring and content iteration don't stop after one delivery. A retainer should buy a defined monthly operating scope, not unlimited access: a fixed strategy and measurement layer, a set number of content or optimization units, an authority allocation if included, and a monthly report. The framing that sells this well: each month buys a defined visibility and content program, with next month's work chosen from what the tracked answers and competitors actually show.
Fixed-price project or sprint. Works when the beginning, end, and acceptance criteria are clear: a paid readiness assessment, a prompt and competitor mapping exercise, a defined content-restructuring project, or a proof of concept. State plainly what happens after delivery. If ongoing monitoring isn't included, say so in the proposal, or the client will assume it is and you'll be doing unpaid work by month two.
Hybrid SEO plus AEO add-on. Often the easiest sell for an agency that already runs the client's SEO retainer. Keep that scope intact and add buyer-prompt research, AI-visibility monitoring, citation analysis, and answer-oriented content restructuring as a distinct, separately reported line item. Don't fold it invisibly into the existing SEO fee. A separate line makes the new service visible, gives it its own success measure, and stops you from absorbing extra strategy and reporting work for free.
Performance or outcome-linked pricing. Handle this one carefully. An agency doesn't control the answer engine, the client's own site changes, third-party sources, or competitor behavior, so a pure performance model for AEO carries real risk. If you use a bonus structure at all, keep a base retainer underneath it, tie the bonus to a precisely defined and auditable change (an agreed improvement in citation coverage for a fixed prompt set, for instance, not "more AI traffic"), and define the baseline, platforms, and observation window up front. Never guarantee inclusion in a generative answer, a specific citation, or a traffic number. No agency controls that outcome, and promising it is the fastest way to lose a renewal.
Common mistake: selling AEO as unlimited "ongoing optimization" with no prompt, page, or revision ceiling. That phrase invites scope creep the moment the client's expectations outrun what you quoted.
Step 5: Price the package from scope and capacity, not from a market average
Good AEO service pricing starts from your own costs, not from a number you saw on a competitor's page. Published pricing guides put small-business AEO programs somewhere around $1,500 to $5,000 a month, mid-market programs at roughly $5,000 to $10,000, and enterprise programs at $10,000 to $20,000 or more, with some full-implementation projects running $25,000 to $100,000. A separate GEO pricing guide places entry programs at $2,000 to $4,000 a month, mid-market at $5,000 to $8,000, and enterprise at $8,000 to $15,000, with a hybrid SEO-plus-GEO add-on running $1,500 to $3,000 for SMB accounts and $4,000 to $6,000 for enterprise ones. These ranges are directional market examples from published agency pages, not a rate card you can hand your client as proof of anything. Use them for context, then build the actual number from your own costs.
When you price AEO services this way, the number comes from your actual labor and margin instead of a guess. For each package, estimate strategy and account-management hours, prompt research and analysis time, content planning and briefing time, editorial review time, technical implementation or coordination, reporting and client-meeting time, and your software and data costs. Add any external authority, PR, or placement costs, then apply the gross margin you actually need. Compare the resulting effective hourly rate against your agency's normal rate before you send the quote. For general context, a 320-respondent Credo survey put the average worldwide SEO-agency rate at $134.14 an hour and the US average at $147.93. A separate Ahrefs survey of 439 providers found an average agency retainer near $3,209 a month, with $100 to $150 the most common hourly band. Again: context, not your quote.
Protect the margin with real boundaries once the price is set. Cap the number of tracked prompts, markets, products, pages, and revision rounds. Separate strategic work from production units so a client can't quietly convert one into more of the other. Decide up front whether your agency or the client implements technical changes, and treat any paid placements or distribution as a pass-through cost unless you deliberately built it into the fee. Review the package after the first cycle and reprice if the evidence shows the scope was too small for what the work actually took.
If you're running the AEO service delivery through a platform rather than manual research and drafting, the software cost is part of this math, not separate from it. DeepSmith's plans run $99, $199, and $399 a month (or $80, $160, and $299 billed annually), scaling from 20 to 90 articles and from ChatGPT-only tracking up to ChatGPT, Perplexity, and Gemini together, with a custom Enterprise tier above that. Whatever tier you're running for a client's account, the client-facing price still needs to cover your strategy, interpretation, editorial review, and margin on top of the software cost. Passing the software price straight through as your package fee leaves nothing for the work you're actually being hired to do.
Step 6: Position AEO next to the SEO retainer, not instead of it
Selling AEO gets easier once you stop implying that traditional SEO is obsolete, because it isn't, and the client's team knows it isn't. The honest comparison is additive: SEO covers keyword research and rankings; AEO adds natural-language buyer prompts and answer coverage. SEO tracks organic traffic and conversions; AEO tracks mentions, citations, share of voice, and which pages actually get cited. On-page optimization becomes answer-first structure, extractable sections, and entity clarity. Link building and authority work extend into citation-worthy assets and source alignment. The monthly SEO report gains a companion: an AI-search visibility report with prompts, platforms, citations, and next actions.
A sales line that captures this without overreaching: SEO helps a client earn organic search demand, and AEO adds a measurement and content layer for the questions buyers now ask inside AI interfaces. You track where they appear, where a competitor appears instead, which sources get cited, and what content can close the gap. Nothing about that claims AEO replaces the SEO foundation. It's the same foundation with a new layer of evidence built on top of it.
Step 7: Pitch the offer to existing and new clients
Everything in the first six steps sets up the actual conversation. Knowing how to sell AEO services still comes down to a good pitch delivered at the right moment, to the right person. For an existing SEO client, the entry point is the relationship you already have, but the value has to be made explicit rather than assumed. Start with the client's own question: which buyer questions do they care about, and do they actually know how they and their competitors show up in AI answers right now? Frame the business risk honestly. It isn't just a ranking that slipped. It's a buyer getting an answer that names a competitor, cites a competitor's page, or describes the client's product incorrectly, with nobody on the client's side even aware it happened.
From there, show the measurement layer you'll add: a defined prompt set, mention rate, citation rate, share of voice, and the pages getting cited. Connect that measurement to the actual work, so the client sees that next month's content comes from prompts being lost and gaps in their own answer coverage, not from a generic content calendar. Offer a bounded first package, whether that's a paid setup, a defined sprint, or a clearly scoped AEO layer added to the existing SEO retainer, and set the expectation window honestly: early signals can show up within weeks, but no platform outcome is guaranteed, and the program should be judged on the leading indicators you agreed to track.
For a new client, lead with their visibility problem rather than your production capability. When a prospect asks an AI engine which vendors to consider, the answer gets built from pages and sources the client isn't monitoring. The pitch is that you'll identify the buyer questions that matter, measure visibility against named competitors, and turn the gaps into an ongoing content and authority plan, while being upfront that you're reporting evidence and probability, not promising a specific outcome no agency actually controls.
This is also where a platform earns its place in the pitch itself, not just the delivery. DeepSmith's Opportunity Agents read a client's visibility and content-coverage data and return ideas with the specific data point that justifies each one, so the backlog you're proposing isn't a brainstormed list, it's evidence a client can check for themselves. And because DeepSmith supports separate workspaces per client, with each account's own brand voice, product facts, and content rules stored in Deep IQ, you can run the same repeatable AEO framework across every account without one client's positioning bleeding into another's drafts.
Common mistake: guaranteeing AI inclusion to close the deal. Generative answers aren't a deterministic ad placement you can promise your way into. Sell the process and the measurement, and let the results speak in the report.
Step 8: Make the report support renewal, not just prove you did the work
The report is part of what you're selling, not paperwork you produce after the fact. A report that only says "we published twelve articles" gives a client nothing to renew against, because it doesn't answer the question they're actually paying to have answered: is this working, and what's next.
A recurring report worth the name includes the prompt set tracked and anything that changed about it, mention rate and citation rate, share of voice against the named competitors, a platform-by-platform breakdown, representative answer changes, which pages are getting cited and by which prompts, which competitor pages or sources are winning the prompts you want, what work got completed this period, and the next actions tied to specific evidence rather than a general sense of momentum. Include the caveats too: platform volatility, sample size, and anything about attribution that could shift the numbers next month.
The renewal conversation follows the same shape. Show which prompts improved, which ones a competitor still owns, which pages or sources are influencing the answers, what the current package addressed, and what the next gap is worth funding. That's a much stronger case for the next invoice than a list of deliverables, because it shows the client you're managing a program toward an outcome, not just producing content on a schedule.
What to do next
If you want the short version of how to productize GEO services without overbuilding on day one: pick one existing client, define a narrow prompt set for them, and price a single bounded package rather than trying to launch the full ladder at once. Use the first reporting cycle to see what the scope actually took to deliver, then adjust the price or the package before you sell it to a second client. If you want to see what a platform behind this kind of program actually produces, from tracked prompts to a finished, linked, publish-ready article, DeepSmith runs a 7-day free trial with real data and real drafts before you pay.
