For Agencies

Running an AEO practice across a book of clients

Launching an AEO practice and operating one are different problems. Launching is a positioning and pricing question. Operating is a question of what you can standardise across a book of clients and what stubbornly cannot be - and the honest answer is that prompt sets, Corporate Context and the judgment calls are per client, while measurement cadence, reporting shape and the action loop are shared. This page is the operating rhythm: what happens weekly, what happens monthly, where a solo operator actually breaks, and what separates a 10-client book from a 25-client one.

Updated 2026-09-18

Questions this guide answers

How do agencies run AEO for multiple clients?
How many AEO clients can one person manage?
What does an agency deliver to AEO clients each month?
How much does an AEO platform cost for an agency?

Direct answer

A multi-client AEO practice works when you are ruthless about which parts are shared and which are not. Shared: measurement cadence, report structure, the action loop, the drafting patterns, and the questions you ask in every onboarding. Per client: the prompt set, the Corporate Context, the competitor list, the measurement market, and every judgment about what to do next.

Get that split wrong in the direction of over-standardising and you ship the same five recommendations to twelve clients, which they notice. Get it wrong in the other direction and every client becomes a bespoke project, which does not scale past about four.

SolCrys prices the agency motion per client organization rather than per seat: Agency at $1,499 a month covers 10 client organizations on ChatGPT, Gemini and Google AI Overviews / AI Mode, and Agency Pro at $4,999 covers 25 and adds Perplexity and Claude to the selectable set. Both carry 30 tracked prompts per client organization; the analysis quotas differ (200 Deep Analyses and 50 Content Audits a month on Agency, 700 and 150 on Agency Pro). Current numbers live on pricing, which is canonical if it and this page ever disagree.

What has to be per client, and why

Three things resist standardisation, and it is worth knowing why before you try.

The prompt set. Two clients in the same category still lose different questions. A prompt set copied across a book measures the agency's assumptions rather than each client's buyers - and since prompt selection determines every chart downstream, a wrong set makes the whole engagement confidently wrong. See how prompt sets are built.
Corporate Context. The approved record of what is true about a company is the thing that stops drafted content inventing capabilities. It is client-specific by definition, it is the slowest part of onboarding, and it is also the part that pays back most - once it exists, drafting for that client gets fast and stays safe.
The measurement market. A client selling in Germany should not be measured from the United States. Market is a per-workspace setting, and changing it starts a new measurement series rather than bending the existing trend line, so it is an onboarding decision rather than an afterthought.

The weekly rhythm

Across a book, the week has a shape. It is deliberately short, because the work that moves numbers is the drafting and the relationships, not the dashboard reading.

Scan the exceptions, not the book. You are looking for movement outside normal engine noise, not for a number on every client. Rolling windows exist so you can ignore single-day moves; watch the 7-day view and let the 30-day view carry the story you tell the client.
Triage into one queue. Findings from audits, citation gaps and market signals all promote into the same action queue with an owner and a due date. One queue across the book, filtered by client, beats twelve separate to-do lists.
Draft against context, review as a team. Drafts are grounded in that client's Corporate Context and go through an explicit review handoff before anything ships - which is also the mechanism that lets a junior operator do the volume while a senior one keeps the standard. The mechanics are in inside the Action Hub.
Close what shipped. Publication and completion are separate states for a reason: the loop closes on re-measurement, not on the day the page went live.

The monthly client deliverable

The report that renews a retainer answers four questions in this order, and only the fourth is a chart.

What changed in how AI answers describe you. Not the score - the substance. Which questions you now win or lose, and who is being recommended instead.

What we shipped, and what it was aimed at. Each action tied to the gap that produced it, which is the difference between a report and an invoice with screenshots attached.

What moved as a result. Re-measurement on the prompts the work targeted, including the ones that did not move. A practice that only reports wins teaches clients to distrust the reporting.

What is next, and why that and not something else. The forward-looking half is where the retainer is actually re-sold.

Where a solo operator breaks

The honest limit is not the dashboard - it is the drafting and the client conversations. Measurement scales almost for free: doubling the book doubles the runs, and nobody reads them all anyway. What does not scale is writing defensible content in a category you do not know and holding a monthly conversation where you are expected to have an opinion.

In practice a single operator can hold the measurement and triage for a sizeable book, and starts to break on drafting volume well before they break on client count. The two useful responses are to narrow the categories you accept - the second client in a vertical costs a fraction of the first - or to split the roles, with one person running measurement and triage across the book and another doing the drafting under review.

That is also the practical difference between the two agency tiers. Ten client organizations is a book one person can genuinely operate. Twenty-five is a team's book, and the tier carries the engine coverage and the analysis quota to match - roughly 28 Deep Analyses and 6 Content Audits per client organization per month, against 20 and 5 on the smaller tier.

What to standardise on day one

Four templates are worth building before the third client, not after the tenth.

An onboarding questionnaire that produces Corporate Context, the competitor list and the measurement market in one pass.
A prompt-review ritual. The set is built from category evidence, but the client has to see it and approve it - both because they know their buyers and because approval is what makes the measurement theirs rather than yours. Cadence guidance: the prompt set lifecycle.
One report template with the four questions above, so the monthly conversation is about the answers rather than about the format.
A definition of done for an action that includes re-measurement. Without it, 'shipped' silently becomes the finish line and the loop never closes.

FAQ

How many AEO clients can one person realistically run?

More than the dashboard suggests and fewer than the pricing implies. Measurement and triage scale well - the platform does the runs and rolling windows let you ignore normal noise. Drafting and the monthly client conversation are what cap you, and they cap you on category knowledge rather than client count. The second client in a vertical costs a fraction of the first, so a narrow book runs much larger than a scattered one.

What do the two agency tiers actually differ on?

Client organizations and depth, not the workflow. Agency at $1,499 a month covers 10 client organizations on ChatGPT, Gemini and Google AI Overviews / AI Mode; Agency Pro at $4,999 covers 25 and adds Perplexity and Claude to the selectable set, with roughly 28 Deep Analyses and 6 Content Audits per client organization per month against 20 and 5 on the smaller tier. Both carry 30 tracked prompts per client organization and neither meters seats. Current numbers are on the pricing page.

Can I reuse one prompt set across clients in the same category?

As a starting point, yes - the category evidence is the same and it saves real time. As the tracked set, no. Two competitors in one category lose different questions, and prompt selection determines every number downstream, so a shared set measures your assumptions rather than each client's buyers. Start from the category set, then replace what does not fit and have the client approve it.

How do I keep quality consistent when a junior operator does the volume?

Two mechanisms do most of the work. Corporate Context bounds what a draft can claim, so the floor does not depend on who wrote it. The explicit review handoff puts a named reviewer on the task before it ships, so the senior person spends their time on judgment rather than on production. The residual risk is prompt selection and the client conversation, which is why those two stay senior.

Free ChatGPT visibility check

Run a free ChatGPT visibility check on a client domain

Start a free workspace with any client domain: 10 buyer-intent prompts through ChatGPT show where the brand is mentioned, cited, or skipped, and who gets recommended instead, plus the first fix to ship.

Start Free

Free · No credit card · About 5 minutes

Related guides