If you’re hiring someone for AI search work, you’re probably going to send out an RFP at some point. The problem is most AEO RFPs I’ve seen are basically SEO RFPs with “ChatGPT” find-and-replaced in. They ask about rankings and how many prompts a tool tracks, and none of that really tells you if a firm can get you mentioned in an answer.
So here’s the one I’d want to get. It’s free, it’s embedded below, and you can grab it as a Word doc.
Download the AEO RFP template (.docx)
Why the old SEO RFP doesn’t fit
A couple of reasons. First, there’s no position one. Ask ChatGPT the same question five times and you can get five different lists, so what you’re really measuring is how often you show up. Any proposal promising a “ranking” in an AI answer is a red flag for me.
Second, a lot of what gets you recommended isn’t on your website. AI leans on what other people say about you, so reviews, press, Reddit and YouTube need to be in scope, not bolted on later. I’ve written about Reddit and YouTube before, and most “strategies” there are garbage.
And third, tracking is cheap now. There are a ton of visibility tools and some are free. If a proposal is mostly a dashboard license, you’re paying for the easy part.
The questions that tell you the most
The full template has 14 questions. If I could only ask a few, it’d be these.
“What would you tell us not to do?” This is my favorite. If a firm can’t name anything, they’re probably going to chase whatever moves the chart that month.
“Show us 10 prompts you’d track for us, and why.” You find out fast who read your RFP and who pasted in a generic list.
“What share of your work is content, technical, and earned media?” There’s no perfect split. But if earned media is zero, I’d want to hear why.
“Who’s actually doing the work?” Names and hours. The people who’ve been doing SEO through a few big Google updates tend to be calmer about all this, and honestly better at it. The pitch team and the delivery team aren’t always the same people.
“Show us a case study that didn’t work.” Everybody has one. The ones who’ll talk about it are usually the ones you can trust with the reporting.
How I’d score it
I put the most weight on the execution plan on purpose. Ethics only gets 5 points, but that’s because it’s really pass/fail. If they won’t sign the guardrails section, they’re out before you score anything.
On reporting, the thing I’d push hardest on is that every monthly report says what they’re going to do next. Tracking AI search isn’t a strategy. A chart by itself doesn’t help you much.
And on guardrails, I’d make them put it in writing. Hidden text for AI crawlers, fake reviews, astroturfed threads, all of it. I went through why that stuff is a bad bet in a field guide to black-hat GEO. If it gets caught, it’s your name on it, not theirs.
The full template
Copy it from here or download the Word version. Swap out anything in [brackets] and delete what doesn’t apply to you.
Request for Proposal: Answer Engine Optimization (AEO / GEO) Services
[Company name] · Issued [date] · Contact: [name, email]
1. Company overview
| Company | [Name, website] |
| What we sell | [One or two sentences] |
| Who buys it | [Customer type, B2B or B2C, buying committee] |
| Markets served | [Countries, regions, or cities] |
| Main competitors | [3 to 5 names] |
| Current SEO setup | [In-house, agency, none] |
| Why now | [AI referrals rising, competitors named and we aren’t, leadership ask] |
2. Goals and how we’ll measure them
Propose targets after your baseline. AI answers change from run to run, so state targets as shares or ranges across repeated runs, not rankings.
| Goal | Measure | 12-month target |
|---|---|---|
| Get recommended | Share of priority prompts where we’re named, by engine | [ ] |
| Be described accurately | Error rate in AI descriptions of our brand, products, pricing | [ ] |
| Get cited | Citations to our pages and to earned sources that mention us | [ ] |
| Beat competitors | Our mention share vs. [competitors] on comparison prompts | [ ] |
| Drive business | AI referral sessions (GA4), assisted conversions, leads or revenue | [ ] |
3. Scope of work
Price each area separately. Mark anything you’d hand to a partner, and anything you think we don’t need.
| Area | What we expect |
|---|---|
| Baseline audit | How each engine describes us and competitors today, across our priority prompts (Appendix A), with repeated runs |
| Prompt research | Real questions buyers ask AI, by stage and persona, including long-tail and comparison prompts |
| Technical access | AI crawler access (robots.txt, CDN and firewall rules), rendering without JavaScript, structured data, sitemaps |
| Owned content | Product, pricing, comparison, FAQ, docs and help pages that AI cites |
| Entity consistency | Same facts about us on our site, schema, Google Business Profile and directories |
| Earned media and PR | Mentions in the third-party sources AI relies on: press, reviews, lists, trade sites |
| Communities and video | Honest, disclosed participation on Reddit, YouTube, forums and review sites (earn it, don’t farm it) |
| Optional: local | Reviews, listings, service-area pages, published pricing |
| Optional: ecommerce | Product feeds and data for AI shopping assistants |
| Optional: developer docs | Docs, llms.txt and code samples that coding agents read |
| Measurement | Tracking, reporting and analysis (section 6) |
| Training | Teaching our team to keep it going after the engagement |
4. What we’ll provide
- Access to [Search Console, Bing Webmaster Tools, GA4, CMS, CRM]
- Current AI visibility data from [tool], if any
- Top questions customers ask sales and support
- Brand, legal and compliance review rules and turnaround times
- A named internal owner, [name, role], plus [X] hours a month from [content / dev / PR]
Tell us what else you need and how much of our team’s time you expect each month.
5. Questions for respondents
300 words or fewer per answer. Specific examples beat general claims.
Approach
- How do you decide which prompts matter for a business like ours? Show 10 you’d track for us and why.
- Walk us through the first 90 days. What do you deliver, and when?
- What share of your work is owned content, technical fixes, and earned media? Why that mix for us?
- How do AI engines decide which sources to trust and cite? What’s your evidence?
- What would you tell us not to do, and why?
Proof
- Share two case studies with a baseline, what you changed, and the result. Include one that didn’t work.
- How do you separate your impact from engine changes, seasonality and our other marketing?
- Provide two client references we can call.
Team and tools
- Who will do the work? Names, roles, hours per month, years in SEO. What’s outsourced?
- Which tracking tools do you use, which do you pay for, and do we get access? Do you own or resell any tool you recommend?
- How do you keep up with engine changes, and how will you tell us when something shifts?
Working together
- What do you need from our team each month, and what happens if we can’t deliver it?
- How do you handle content review in a [regulated / brand-sensitive] business?
- What does the handoff look like if we part ways? What do we keep?
6. Measurement and reporting
- Track the agreed prompt set across [ChatGPT, Google AI Overviews / AI Mode, Gemini, Perplexity, Copilot, Claude] at least [weekly / monthly].
- Run each prompt multiple times. Report shares and ranges, with the method written down.
- Report mentions, citations (which URLs), sentiment and accuracy separately.
- Tie visibility to GA4 AI referral traffic and to leads or revenue where possible.
- Monthly: a short written summary, actions taken, and next month’s plan.
- Quarterly: a review against section 2 targets and a reset of the prompt set.
- We own all data, prompt sets and content, and can export them any time.
7. Ethics and guardrails
Confirm in writing that you won’t do any of the following on our behalf:
- Hidden text, cloaking, or instructions aimed at AI crawlers that people can’t see
- Fake reviews or accounts, astroturfed Reddit or forum threads, undisclosed paid posts
- Mass-produced, near-duplicate pages built only to catch prompts
- Negative content or misinformation about competitors
- Guaranteed placements or “guaranteed #1 in ChatGPT” claims
List any paid placements, sponsored lists or affiliate relationships you’d use, and how they’re disclosed.
8. Timeline and submission
| RFP sent | [Date] |
| Questions due | [+1 week] |
| Answers shared with all respondents | [+10 days] |
| Proposals due | [+3 weeks] |
| Finalist presentations | [+4 weeks] |
| Decision | [+5 weeks] |
| Engagement starts | [Date] |
Send a PDF of no more than [15] pages plus a separate pricing sheet to [email].
9. Evaluation and scoring
Each criterion is scored 1 to 5 and multiplied by its weight, for a total out of 500.
| Criterion | Weight | What a 5 looks like |
|---|---|---|
| Execution plan | 25 | Clear 90-day plan; most of the budget goes to doing the work |
| Strategy and fit | 20 | Prompts and priorities specific to our buyers |
| Measurement rigor | 15 | Repeated runs, stated method, honest about variance |
| Proof and references | 15 | Case studies with baselines, including a miss |
| Team and SEO experience | 10 | Named senior people who’ve been through past search changes |
| Price and value | 10 | Itemized; price tied to work, not prompt counts |
| Ethics and transparency | 5 | Signs section 7; discloses tool and partner ties |
Red flags: guaranteed AI rankings, a proposal that’s mostly a dashboard license, no named team, no answer to “what wouldn’t you do?”, long contracts with no exit.
10. Budget and terms
- Budget range: [$X to $Y per month / per project]
- Audit priced as a fixed fee, separate from ongoing work
- Monthly retainer priced by scope area, with hours or deliverables
- Tool costs listed separately
- Term: [6] months, then month to month with [30] days’ notice
- We own all content, data, prompt sets and reports
Appendix A: Priority prompts
10 to 30 questions buyers ask before choosing a company like ours, pulled from sales calls, support tickets and site search.
| Prompt (as a buyer would ask it) | Stage | Who’s named today |
|---|---|---|
| [What’s the best [category] for [use case]?] | Comparison | [ ] |
| [[Brand] vs [competitor] for [need]?] | Decision | [ ] |
| [How much does [service] cost in [city]?] | Research | [ ] |
| [Is [brand] good for [situation]?] | Decision | [ ] |
One tip before you send it: fill in Appendix A yourself, even roughly. Your own list of prompts is the quickest way to tell who read the RFP. The good firms will add to it. The others will ignore it.
Want us on your RFP list, or help writing one? See our services or get in touch.
Leave a Reply