AEO agency SOPs only work if the prompt set stays frozen. Every other procedure in an answer engine service rests on that one. If someone edits the questions in place, last month's numbers and this month's numbers were collected under different conditions, and the trend line you show the client means nothing. So the first AEO SOP to write is the one that versions the prompt set.
AEO agency SOPs are the written procedures that make answer engine optimization repeatable: how a prompt set is built, how answers are sampled, how an answer-first page is checked and how results are reported. The seven-field structure for any SOP is in SEO agency SOPs. The operating model these sit inside, including how to size the sampling hours, is in AEO agency operations.
Why editing a prompt set in place wrecks the trend
I have seen an agency's reporting where the question list changed from one month to the next and the visibility figure moved with it. A report like that cannot say whether visibility rose or the questions got easier. That is the whole reason to freeze the list.
A version log makes the change visible and keeps the old data usable. Each entry records the version, the date, what changed, who approved it and whether a new baseline was taken.
The rule is simple. A change creates a new version and a new baseline, and the old trend ends cleanly. Nobody edits version one.
The AEO agency SOPs to write, and what each is for
| SOP | One-line purpose |
|---|---|
| Prompt set build and versioning | Collect, group, approve and freeze the question set |
| Baseline sampling | The first measurement run, before any change ships |
| Citation sampling run | The repeating measurement, with fixed prompts, engines and run counts |
| Answer accuracy review | A person checks what engines say about the client |
| Answer-first page brief | What a page must specify before it is written |
| Answer page QA | The gate a page passes before it ships |
| Structured data validation | How markup is specified, tested and released |
| AEO report | How sampled results become a client report |
Intake, access and scope-change SOPs from your SEO set carry over, with prompt groups added to the scope document. The sampling run is covered in the operations post as a capacity item, so I will not write it out a third time here. In my team's setup the run is a fixed question list on a schedule through a professional AI-visibility tool, read alongside Search Console, analytics and Google Analytics traffic from LLM referrals.
Worked SOP 1: Prompt set build and versioning
Trigger: AEO scope signed. Owner: the AEO lead. Output: an approved, dated prompt set.
- Ask the client for the questions buyers ask before, during and after a purchase decision, using real wording from sales calls or support tickets where it exists.
- Draft each prompt in natural language, the way a person would type it to an answer engine.
- Group prompts by intent: category, comparison, problem and brand-specific.
- Remove duplicates that differ only in wording, but keep two phrasings of the most important prompts.
- Record which engines matter to this client, and any location or language the prompts must be tested under.
- Send the set to the client for approval and record who approved it and when.
- Freeze it. Any change goes through the version log above.
The check: a second team member confirms every prompt reads as a question a buyer would ask, and not as an internal term. The set does not enter sampling until the approval is on record.
Worked SOP 2: Answer page QA that the writer runs first
Most of what a reviewer corrects on an answer page can be written as a rule, so the writer should check it first. The writer runs every line below on their own draft, and the reviewer only sees pages that have passed.
| Check | Passes when | Mechanical or judgment? |
|---|---|---|
| Opening | The first two sentences answer the target question on their own | Mechanical: read them in isolation |
| Specificity | The answer states something concrete, with no vague opener | Judgment, but the writer flags doubts |
| Claims | Each factual claim has a source, a date or a named basis | Mechanical |
| Brand facts | Names, locations and offerings match the approved brand sheet | Mechanical |
| Headings | Phrased as questions or topics buyers use | Mechanical |
| Structured data | Validates and matches what is visible on the page | Mechanical |
| Crawl and links | Crawlable, indexable and linked from relevant pages | Mechanical |
| After release | Page is added to the next sampling run for its prompt group | Mechanical |
Most of the lines are mechanical, which frees the reviewer for the parts that are not: whether the opening is actually good and whether the claims are true and current. A page with an unverified claim does not ship, and a page that skips the last line never enters sampling, so nobody learns whether it worked.
Writing steps that survive tool changes
Write each step as an action and a record, because tools change. State the limits of the method in every sampling record too: answers vary between runs, and measurement in this area is still maturing.
If you are weighing how many procedures your team can realistically maintain, that is a capacity question I help with through AEO agency utilization consulting. The strategy behind the procedures is in the AI search optimization guide, and the entity-side procedures are in GEO agency SOPs.
FAQ
What SOPs does an AEO service need?
At minimum: prompt set build and versioning, baseline sampling, a repeating citation sampling run, answer accuracy review, answer-first page brief, answer page QA, structured data validation and an AEO report. Intake and scope-change SOPs from your SEO set carry over with prompt groups added to the scope document.
Why version the prompt set instead of editing it?
Editing prompts in place breaks comparisons over time, because two months of data were gathered under different questions. A new version starts a new baseline and closes the old trend cleanly. Record who approved each version and when, so a change in results is never confused with a change in the questions.
How is an AEO SOP different from an SEO SOP?
The structure is the same, but the inputs and checks differ. AEO SOPs start from a frozen prompt set, measure by sampling answers repeatedly across engines, and QA pages by whether the opening sentences answer a question on their own. SEO SOPs start from keywords and check rankings and technical correctness.
Who should run the answer page checklist, the writer or the reviewer?
The writer should run it first on their own draft, so the reviewer only sees pages that have already passed the mechanical lines. That leaves reviewer time for judgment: whether the opening is good and whether claims are true and current. It also stops the same avoidable corrections bouncing back repeatedly.
