Selling AEO Without Overclaiming
The consultants who will still have AEO clients in two years are the ones scoping it honestly now. Here's how to price work whose results arrive slowly, what to promise, and how to report a quarter where the number didn't move.
Every new marketing discipline goes through a phase where the loudest practitioners make claims the mechanism cannot support, buyers get burned, and the field spends years recovering trust it did not need to lose. SEO went through it. Social went through it. AEO is in it now.
For an independent consultant this is both a hazard and an opening. The hazard is competing against people promising guaranteed citations in thirty days. The opening is that those promises produce churned clients within two quarters, and a practice built on honest scoping inherits them.
This is about how to do the second thing without losing every deal to the first.
What you cannot promise, and why saying so wins
Four things are outside anyone’s control, and a client will eventually discover each of them. Discovering them from you in the first meeting is a credential; discovering them in month four is a refund conversation.
You cannot guarantee a citation or a position. Engines are non-deterministic — the same prompt run twice returns different brands with nothing having changed. Anyone guaranteeing placement is guaranteeing the output of a stochastic system they do not control.
You cannot edit a model. If an engine believes something false about a client, there is no correction form. You can change what live retrieval finds and, slowly, what the third-party consensus says. You cannot reach the weights.
You cannot attribute revenue to a mention. The chain from an AI answer to a purchase routinely runs through a direct visit weeks later with no identifier surviving. A modelled revenue figure will not survive a CFO asking about its assumptions.
You cannot make it fast. Grounded answers respond in days to weeks, third-party consensus over months, training-derived knowledge across model generations. How long does AEO take is the version worth sending a prospect before they sign.
The counterintuitive part is that stating these improves close rates rather than hurting them, for a specific reason: your buyer has usually already heard the guaranteed-placement pitch and found it slightly implausible. Being the person who explains why it is implausible converts scepticism into a reason to trust you. What AEO cannot do is the long-form version, and sending it to a prospect is a stronger sales asset than a case study.
How to scope it
Three shapes, and mixing them up is where consultants lose money.
The audit is a project, not a retainer. Technical accessibility, entity resolution, extractability, a baseline measurement, a prioritised backlog. It has a defined end, it produces the fastest visible wins, and it is genuinely valuable. It also finishes — building a recurring engagement on it means either padding it or watching the retainer die in month three.
The programme is the retainer. Third-party presence, original data, content restructuring, sustained monitoring, quarterly reassessment. This is the actual service and it is a harder sell because its results are slow and partly outside your control.
Diagnosis is billable and should be. Two clients with identical symptoms — ranking on Google, absent from AI — can have entirely different causes: an entity collision, a rendering problem, no third-party footprint. Working out which is skilled work, it eliminates most of the possible causes at once, and consultants routinely give it away as pre-sales. When Google ranks you and AI doesn’t covers the diagnostic frame.
The pricing implication: charge properly for the audit, price the programme on a horizon the mechanism can actually deliver against, and do not promise the programme’s outcomes on the audit’s timeline.
The quarter where nothing moves
This will happen, and how you handle it determines whether the engagement survives.
A quarter can produce no visible movement for four legitimate reasons. The work was upstream — entity clarity, third-party groundwork — and has not surfaced yet. The client did not execute their half. A model update moved the whole category. Or the change was real and smaller than the sampling noise.
The way to survive it is to have made the report structure honest before you needed it:
Report activity monthly, outcome quarterly. A monthly outcome report shows noise, and explaining variance every month teaches a client that the service does nothing.
Show the leading indicators. Sources engines cite in the category, pages now reachable that were not, third-party mentions earned, prompts newly covered. These move before presence does, and they are evidence of work rather than of luck.
Say when a change is inside the noise. A mention rate over a hundred observations carries roughly a ten-point margin. Calling a five-point rise a win means calling a five-point fall a loss next quarter. How many runs before you trust an AI visibility number is worth having in an appendix the first time it comes up.
Name model updates when they happen, and flag the possibility in month one. Saying it early costs nothing. Saying it the day it happens sounds exactly like an excuse.
Where the real leverage is
For a solo practice, the highest-value thing you can offer is not execution capacity — you do not have much — it is judgement about sequencing.
Most clients arrive wanting the visible, external work: press, citations, presence. Most of them need the cheap internal work first, because their pages do not state their claims plainly, their entity is ambiguous, or a crawler cannot render their site. Doing the external work first amplifies a muddled message and costs the client considerably more.
The GEO paper is useful evidence here, and it is worth quoting to clients: adding quotations, statistics and cited sources to a page measurably raised its visibility in generative engines, while keyword-oriented edits did not. That is a controlled result pointing at the client’s own content as the cheap lever, and it is a more persuasive argument for sequencing than an assertion of experience.
What to do about competitors who overclaim
Not much, and specifically not this: do not spend a sales conversation attacking them. It reads as insecurity and it centres them.
The durable move is to give the buyer a tool rather than an opinion. How to audit an AI visibility vendor’s numbers is a list of methodology questions any buyer can ask anyone — are these real engines or simulated, is the answer grounded, is a “citation” a link or a domain scraped from prose, how many samples. Hand it over and invite them to ask you the same questions.
A prospect who asks a guaranteed-placement vendor how many samples underlie their score does your argument for you, and you have not said a word about a competitor.
The counter-argument
The strongest objection: this is a comfortable philosophy for someone with an established reputation, and a consultant who needs three clients this quarter cannot afford to lead with everything that will not work while a competitor promises results in thirty days.
That pressure is real and the answer is not to pretend it away. Two things make honest scoping commercially viable rather than merely admirable.
The first is that the audit is genuinely fast and genuinely valuable. You can deliver something concrete in two weeks — a rendering problem found, an entity collision identified, a baseline nobody had — and that is a real, sellable, quick win that does not require promising an outcome you cannot control.
The second is that the overclaiming competitor’s clients become available. A guaranteed-citation engagement fails on a predictable schedule, and the client emerges knowing exactly which questions to ask next time. Being findable and credible when that happens is a slower business development strategy than overclaiming, and it compounds where the other does not.
None of which means it is easy. It means the trade is between a faster start and a practice that still exists in three years.
Where to start
- Write down your limits and put them in the proposal, not the small print.
- Price the audit as a project and the programme as a programme.
- Set the cadence at kickoff: activity monthly, outcome quarterly.
- Give prospects the vendor-audit questions and invite them at you.
AEO for consultants covers the practice shape, and is AEO worth it is a reasonable thing to send a prospect who is not sure yet — including the parts that argue against.
Written by
Team @ LLM MetrixWe research and write about AI brand visibility, GEO, AEO, and the evolving AI search landscape.
See how your brand appears in AI search
Track your visibility score across ChatGPT, Claude, Gemini, Perplexity, and more — free to start.
