AEO RFP Template: 28 Questions to Evaluate an Agency Honestly

Why Most AEO RFPs Fail to Surface Real Capability

The standard procurement RFP, ported from digital marketing or SEO templates, does not work for AEO. The questions are too generic. The scoring rubrics reward marketing language over substance. The result is that procurement teams routinely shortlist agencies based on response polish rather than measured capability, then discover the gap six months in.

A good AEO RFP does three things differently. It asks specific operational questions whose answers cannot be faked by a strong proposal writer. It requires evidence in the form of artifacts, prompt sets, or methodology documents rather than claims. It includes a working session component where the agency has to demonstrate capability in real time on the buyer's actual content.

The 28 questions below have been refined across roughly sixty AEO RFP processes over the last two years. They are not exhaustive. They are the minimum set that reliably produces a confident shortlist. Each question is paired with the answer pattern that indicates strong capability, so a procurement lead without deep AEO expertise can still score responses defensibly.

Bucket One: Measurement and Attribution (Questions 1 to 5)

Measurement is the foundation. An agency that cannot measure rigorously cannot prove their work is moving the metrics that matter. Spend the most evaluation energy here.

Question 1 asks the agency to walk through their citation share measurement methodology, including the AI assistants covered, the prompt set construction, and the refresh cadence. Strong answers include specific assistants by name (typically ChatGPT, Perplexity, Claude, Gemini, sometimes Copilot and DeepSeek), describe how prompts are sampled or selected, and disclose whether measurement is live API based or scraping based.

Question 2 asks how attribution from AI assistants to website sessions is handled. Strong answers acknowledge the limits of referrer data, describe whatever attribution model they use, and openly admit the precision range. Question 3 asks for a sample monthly report from a current client (with brand details redacted) to evaluate reporting depth. Question 4 asks whether the prompt set is shared with the client and whether the client can modify it quarterly. The only acceptable answer is yes to both. Question 5 asks how the agency reports declines as well as gains.

Bucket Two: Technical Capability (Questions 6 to 10)

AEO is half technical. Agencies that lean only editorial cannot fix the structural reasons content is not being cited.

Question 6 asks the agency to explain how they think about robots.txt and AI bot access in 2026. Strong answers have a clear point of view, distinguish between training bots and retrieval bots, and reference specific bot user agents. Question 7 asks for their schema markup standards across Article, FAQPage, HowTo, Organization, and Product schemas as applicable to your business. Strong answers reference Schema.org specifications, describe a JSON-LD pattern library, and explain how schema decisions are tested.

Question 8 asks how they approach content freshness signals and lastmod handling. Question 9 asks how they evaluate which AI assistants currently crawl the brand's site, including specific tools or methods used to detect crawl behavior. Question 10 asks for a recent technical audit deliverable from a current client, redacted. The deliverable should be specific to AEO concerns rather than a generic SEO audit with AEO branding pasted on top.

Bucket Three: Content Capability (Questions 11 to 15)

Question 11 asks how article topics are selected and prioritized. Strong answers describe a method that combines citation gap analysis, competitor coverage mapping, persona prompt research, and authority profile considerations. Weak answers reference a proprietary topic engine without explaining the inputs.

Question 12 asks about content production rate and team composition. The right answer depends on the engagement size, but the question reveals whether the agency uses generative tools with light editorial review (high volume, low rigor) or research first writing with substantial editorial time (lower volume, higher rigor). Question 13 asks how they handle the humanization of AI generated drafts. Strong answers describe a structured editorial pass with specific patterns to remove and brand voice profiles being built per client.

Question 14 asks for two recent content pieces from current clients that drove the largest measured citation lift, with the lift documented. Question 15 asks how they balance net new content production with optimization of existing content. The right ratio shifts over time but agencies that only do one or the other are missing half the discipline.

Bucket Four: Governance and Risk (Questions 16 to 19)

Question 16 asks how the agency handles brand safety considerations when generating content. Question 17 asks about their factual accuracy review process, particularly for any statistics, competitor claims, or product specifications. Strong answers describe a multi step verification process and a clear position on what counts as a verifiable source.

Question 18 asks how they handle content that mentions competitors. Strong answers commit to verifiable facts only and explicit refusal to fabricate competitor weaknesses. Question 19 asks about their position on AI generated imagery and visual content, including disclosure practices and brand guideline integration.

Bucket Five: Results and References (Questions 20 to 23)

Question 20 asks for three reference clients of similar size and industry, with named contacts the buyer can speak to. The agency should be willing to facilitate the reference calls within two weeks. Question 21 asks for a documented case study showing citation share trajectory from month one to month twelve with at least one current client. Strong case studies show the trajectory with monthly data points, not just before and after.

Question 22 asks about clients who have churned in the last twelve months and the reasons for churn. Every agency has churn. Agencies that claim zero churn are either very small or not being honest. Question 23 asks for the agency's worst quarter with any current client, what happened, and what they did about it. The willingness to answer this question candidly is itself a strong signal.

Bucket Six: Team and Process (Questions 24 to 26)

Question 24 asks about the team composition that will be assigned to the account, including names, roles, and relevant experience. Pitch teams and delivery teams should overlap meaningfully. Question 25 asks about the agency's average tenure of the AEO practice leads. Long tenure correlates with deeper accumulated expertise and lower delivery risk.

Question 26 asks about the working rhythm: how often the client sees the team, what the QBR cadence looks like, and how escalations are handled. Strong answers describe a defined weekly cadence, monthly written updates, quarterly business reviews, and a named escalation contact above the account team.

Bucket Seven: Commercials and Terms (Questions 27 to 28)

Question 27 asks for the fee structure with explicit detail on what is included and what is treated as out of scope. Watch for vague scope language that creates room for change orders. Question 28 asks about contract terms including the minimum commitment, the termination notice period, and the handling of work product on termination. Twelve month minimums are reasonable. Thirty day termination notice with full work product transfer is standard.

This is where OnlyAEO publishes pricing and terms openly during RFPs, including the prompt set methodology, the team composition that would actually deliver, and a documented sample monthly report. Buyers running competitive RFPs are welcome to use the 28 question framework against us as one of the shortlisted agencies. Honest evaluation is good for the discipline and good for buyers, and it is the right way for a procurement team to surface real capability before signing a twelve month commitment.