Run the AEO retainer from first pitch to renewal
AEO is the rare new line item clients ask about before agencies pitch it. The obstacle is rarely demand. It is having a repeatable audit, a defensible baseline, and a report that renews the retainer without a bespoke analysis every month.
This page describes how agencies structure the practice: what goes into a prospect audit, how to scope the work into packages a client can stage, how to run the same methodology across a book of clients, and what the monthly report contains. Partner terms, reporting arrangements and programme specifics live on /agencies; this page covers the delivery method.
Why agencies build the practice this way
The audit is the pitch
A baseline showing a prospect absent from their own category answers, with a competitor named three times and the sources that decided it, is more persuasive than any deck. It takes minutes to produce and the prospect can verify it by asking the same question. Findings a prospect can reproduce convert; general warnings about AI search do not. The technical findings, such as a blocked retrieval agent or a client-rendered pricing page, are equally verifiable and give the engagement an immediate first deliverable.
One workflow across every client
The same crawler-access checklist, the same rendering test, the same prompt-set methodology and the same report shape, whatever the client's category. That is what lets the practice scale past the one person who understands it, lets junior staff run the audit, and makes reporting a process rather than a monthly essay. The categories differ; the method does not. AI Search Monitoring at /features/ai-search-monitoring is the tracking layer that runs across the book.
Renewals need a trend line
Mention rate and citation share on a fixed prompt set give a month-over-month number that justifies the retainer without re-arguing the strategy. The cited-source list gives the next month's work plan, so the report always ends with what happens next. Clients renew on visible progress and a clear plan; they churn when each month's report is a fresh argument for why the channel matters at all.
It extends the retainers you already sell
AEO shares its technical foundation with SEO and its source work with PR, which means most agencies sell it as a workstream inside an existing retainer rather than a new practice. The SEO team already runs crawl audits and structured data; the PR team already pitches the sites that turn out to be cited. What is new is the crawler list, the extraction test and the metric set, and those are learnable in a week with /learn/aeo-101.
How the agency workflow runs
From prospect audit to renewal reporting, repeatable across a client book.
- 1
Audit the prospect
Check the served robots.txt and bot-management rules for OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User and ClaudeBot, separately from training crawlers such as GPTBot and CCBot. Fetch key templates without JavaScript and confirm body copy and JSON-LD are present. Validate structured data with /tools/json-ld-validator. Then run the prospect's category, comparison and problem prompts across the main engines and record who is named and which sources are cited. Put the findings the prospect can verify themselves on the first page.
- 2
Scope from the gaps
Split the findings into three work packages with different costs and timelines: technical fixes (access, rendering, structured data, llms.txt), on-site restructuring (answer-first editing, consolidation, comparison and FAQ content) and off-site source work (review profiles, roundups, community). Estimate each separately so the client can stage the spend and so early wins from the technical package are not held hostage to the slower content work. Put the monthly monitoring and reporting retainer on top as the recurring line.
- 3
Fix the mechanics first
Deliver the technical package in the first weeks: allow-list the retrieval agents, get key templates server-rendering, complete Organization, Article, FAQPage and Product markup, and publish an llms.txt from /tools/llm-txt-generator. These are fast, verifiable and produce the first movement in citation share, which buys time and trust for the slower work. Re-run the prompt set after shipping so the report can show a before and after on the same questions.
- 4
Run the content and source programme
On-site, apply the extraction test to existing pages before writing new ones: direct answer first, one claim per passage, consolidate overlapping pages, add definitions. Off-site, take the cited-source list from the baseline and work it: complete review profiles, pitch the roundups with facts, engage in the community threads where the client is discussed. Reddit monitoring at /reddit-monitoring shows which threads are being cited. The PR and content teams already have these skills; the source list gives them the targets.
- 5
Report on a fixed cadence
Run the same prompt set every month across the same engines and report mention rate, citation share and competitor share per engine, with the cited-source list, the month's actions and the next actions. Keep the prompt set stable; changing it resets the trend. Add recovered AI-referred traffic from /features/ai-traffic-decoder where the client's analytics allow it, presented as a floor. The report should take an hour to assemble, not a day.
- 6
Standardise across the book
Turn the audit into a checklist, the prompt-set method into a template per category type, and the report into a fixed layout. Onboard delivery staff with /learn/aeo-101 so the language and the checks are consistent. Review the method quarterly as engines change their crawlers and citation behaviour, and roll changes to every client at once. That is what makes the practice a product rather than a specialist's side project.
How agencies use it
New business audits
A concrete, verifiable finding in the pitch, such as a blocked retrieval agent or a competitor named three times in the prospect's category answers, rather than a general warning about AI search. It shortens the sales cycle because the prospect can check it themselves.
Retainer expansion
Adding an AEO workstream to an existing SEO or PR retainer, scoped as a technical package plus monthly monitoring. The technical foundation and the source work overlap with what the team already does, so the margin is in method rather than headcount.
Multi-client reporting
One prompt-set methodology and one report layout across the book, so the monthly cycle is a process a coordinator can run. The per-client differences are the prompts and the cited sources, not the format.
Team enablement
A documented workflow, a shared checklist and a common vocabulary mean the practice is not trapped in one specialist. New staff can run an audit in their first month, and the agency can sell the service without the founder on every call.
Migration and launch safeguards
Checking a client's replatform or site launch for retrieval-agent access and server rendering before it ships. Answer engine visibility is easy to lose in a migration, and catching it in staging is a retainer-justifying save.
What to tell clients about each engine
Clients ask about ChatGPT by name; the report should cover all of them, and the audit checks differ slightly for each.
ChatGPT
The engine most clients ask about by name. Search-style answers fetch live through OAI-SearchBot and ChatGPT-User, so the access and rendering checks in the audit apply directly, and GPTBot is a separate training decision. Citations show as links, which makes citation share reportable. Category answers lean heavily on review and roundup sources, so the off-site package is usually where the ChatGPT result is decided.
Perplexity
The fastest feedback loop for reporting. Sources are shown prominently, it favours recently updated pages, and PerplexityBot and Perplexity-User both need access. It is often the first engine where a client's citation share moves after the technical package ships, which makes it useful for early proof in the first quarterly review.
Google AI Overviews and AI Mode
Both draw on the Google index via Googlebot, so the client's existing crawlability carries over and Google-Extended is not the switch. Selection favours pages that answer sub-questions directly, which rewards the answer-first restructuring package. Clients already reporting on Search Console find this the easiest engine to explain, and the overview can cite pages that do not rank.
Gemini
Grounds on Google Search and is the default assistant in Google Workspace, so it matters for clients whose buyers work there. Google-Extended governs training and grounding use beyond search, which is the trade-off to explain if a client wants to block it. Entity clarity, meaning a consistent Organization record with sameAs, affects how Gemini describes the client against competitors.
Microsoft Copilot
Grounds on the Bing index, which many client sites have never been tuned for. Bingbot access, Bing Webmaster Tools verification and IndexNow submission are the audit items, and they are quick wins for clients selling into Microsoft-centric enterprises. Include it in the prompt set from the start rather than adding it later, so the trend line is complete.
The numbers on the monthly client report
Four client-facing metrics and one internal one that tells you whether the practice is scaling.
- Mention rate per client
- The share of the client's prompt set where they are named, per engine. It is the headline number on the monthly report and the one clients remember. Present it against the previous month and the baseline, not week to week, because movement is lumpy.
- Citation share
- Of the sources each engine cites for the client's prompts, the share that are the client's pages. It responds first to the technical and on-site packages, so it is the early-proof metric for the first quarter and the evidence that the fixes worked.
- Cited-source coverage
- The share of cited sources on which the client is present and current. It is the metric for the off-site package and the leading indicator for mention rate, and it turns the source list into a number the client can watch move.
- Report turnaround
- An internal metric worth tracking: hours from data pull to report sent, per client. If it grows, the method has drifted into bespoke analysis and the practice will not scale. The target is an hour of assembly and a short narrative.
Mistakes agencies make when building the practice
Mistake · Pitching AI search with generic statistics instead of the prospect's own data.
Fix · Run the prospect's category prompts and show who is named and which sources decided it. A finding they can verify themselves converts; a slide about market trends does not.
Mistake · Scoping the whole engagement as one undifferentiated block.
Fix · Split it into technical, on-site and off-site packages plus monitoring, priced separately. Clients can stage the spend, early technical wins land on time, and the slower source work is not blamed for a late first deliverable.
Mistake · Changing the prompt set every month to chase interesting questions.
Fix · Fix the set at scoping, add to it only at quarter boundaries, and report the original set throughout. A moving prompt set has no trend line, and the trend line is what renews the retainer.
Mistake · Reporting recovered AI traffic as if it were the full value of the channel.
Fix · Present it as a floor next to mention rate and citation share. Most of the influence happens in conversations that never produce a click, and a client taught to judge the retainer on referrals alone will undervalue it.
A worked example: a twelve-person SEO and content agency
A hypothetical twelve-person SEO and content agency with a book of mostly B2B and D2C clients gets asked by three clients in one quarter whether they should be worried about ChatGPT. The agency builds a prospect audit from the checklist on this page and runs it on its own clients first. Two of them turn out to block PerplexityBot at the CDN, and one D2C client's product pages render price and stock client-side.
The agency packages the findings as a technical fix package, ships it within the month, and re-runs each client's prompt set. Citation share on Perplexity moves for both of the previously blocked clients, which becomes the first slide of the AEO upsell to the rest of the book. The upsell is scoped as technical package, on-site package, off-site package and a monthly monitoring line, with the client choosing which packages to stage.
For delivery, the SEO lead owns technical and on-site, the content lead owns off-site, and an account coordinator runs the monthly report from a fixed layout: mention rate, citation share and competitor share per engine, the cited-source list, actions taken and actions next. The prompt sets are fixed per client and revisited quarterly. New staff take a short course before touching a client account.
Six months on, the AEO line sits inside most of the agency's SEO retainers rather than as a separate sale. The report takes an hour per client to assemble, the D2C clients get a Product schema and feed module in their checklist, and the agency's own pitch now opens with the prospect's baseline rather than a slide about AI search. Partner and programme details for running this on the platform are on /agencies.
Frequently asked questions
How do we price an AEO retainer?+
Most agencies scope it as a bounded audit-and-fix engagement followed by a monthly monitoring and content retainer. The technical audit is estimable in hours; the ongoing content and source work is where the recurring value sits. Pricing is the agency's own decision; partner options and programme details are on /agencies.
Can we white-label the reporting?+
Reporting presentation and partner arrangements are handled through the agency programme, and the current options are described on /agencies. The methodology on this page is the same regardless of how the report is branded, so the delivery workflow does not change with the arrangement.
How is this different from the SEO work we already sell?+
It shares the technical foundation but adds AI-specific crawler access, extraction-oriented content structure, off-site source work and a metric set based on mentions and citations rather than rankings. Most agencies sell it as an adjacent workstream inside an SEO or PR retainer rather than as a replacement.
What do we tell a client who asks whether this is worth it?+
Show them their own baseline. A client who sees a competitor named three times in their category answers while they are absent, and sees which sources decided it, generally does not need the argument made abstractly. Pair it with one technical finding they can verify themselves.
How many prompts should a client's set contain?+
Enough to cover the category, comparison, use-case and problem questions without becoming noisy; for most clients that is thirty to fifty per market. Build it from sales calls, support tickets and community threads rather than keyword tools, fix it, and only add to it at quarter boundaries so the trend line holds.
How do we run this for clients in very different industries?+
The method is the same; the prompts and the cited sources differ. Retail clients need Product schema and feed checks and lean on review roundups; SaaS clients need documentation and pricing visibility and lean on review platforms; local and services clients lean on directories and community. Keep one checklist with industry modules.
What if a client blocks AI crawlers for policy reasons?+
Separate the decision into training and retrieval. Blocking GPTBot, CCBot and Google-Extended is a defensible policy choice. Blocking OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User or ClaudeBot removes the client from answers. Present the trade-off with the baseline data and let the client choose with the consequence visible.
How quickly can we show a client results?+
Technical fixes register on engines that fetch live within days, and citation share on Perplexity usually moves first. On-site restructuring shows over weeks. Off-site source work is the slowest, because it depends on third parties. Set that expectation at scoping, and report the early technical wins while the slower work runs.
Explore the features behind this solution
Go deeper
Add AEO to the retainer
Run a prospect audit, baseline their category, and open the conversation with a finding they can verify.