What Should Be Included in an AEO Agency RFP for a B2B Software Company?
An AEO agency RFP for a B2B software company should do three things. It should give agencies enough context to propose sensibly: your product, buyer segments, competitors, and current SEO and content position. It should set measurement requirements precisely, because AI answers vary and a loose brief invites flattering numbers. And it should make agencies show their method and evidence rather than promise outcomes. In practice that means sections on background, objectives and scope, measurement, method, team and case evidence, security and data handling, commercials, evaluation criteria, and a small paid test. The measurement section does the most to separate strong agencies from weak ones.
What should each section contain?
1. Background
- What the product does, in two or three sentences, and the categories buyers put it in.
- Buyer segments: company size, industries, roles involved in the purchase, and regions.
- Main competitors, and the ones you lose to most often.
- Current position: organic traffic trend, content volume, review platform presence, and any AI visibility data you already have.
- Tools and teams already in place: SEO platform, content team, PR agency, product marketing.
2. Objectives and scope
- Business objective: for example, appearing on shortlists for mid-market buyers in two categories, or correcting outdated descriptions of the product.
- Engines: ChatGPT, Google's AI Overviews and AI Mode, Gemini, Perplexity, Copilot and Claude, or the subset your buyers use.
- Markets and languages.
- Work in scope: measurement, content, technical access for AI crawlers, comparison pages, review programmes, digital PR, or a subset.
- Work out of scope, and which in-house teams the agency must work with.
3. Measurement requirements
This section should be specific:
- Question set: the agency proposes one, you approve it. Require a split between unbranded questions (category, problem, comparison) and branded questions, reported separately.
- Runs: how many times each question is run per engine, and how often.
- Evidence: full answer text and cited sources stored for every run, and available to you.
- Data ownership: you own the question set and all collected data, with export in a standard format.
- Metrics: mention rate on unbranded questions, position, description accuracy, and citation share for key sources. Ask how each is calculated.
- Changes: how the agency handles model updates and question set changes without breaking trend lines.
4. Method
Ask agencies to explain, in their own words:
- how they decide what to change first;
- how they tell a real change from run-to-run variation;
- how they link a change in answers to specific work;
- what they would do in the first 90 days for a company like yours.
5. Team and evidence
- Who will work on the account, their experience, and how much time each will spend.
- Two or three case studies from B2B software, with the question set, the method, the before and after figures, and the time frame.
- References you can call.
Be sceptical of case studies built only on branded questions, which inflate results. See whether brand-name prompts inflate AI visibility scores.
6. Security and data handling
- Which tools and AI providers the agency uses, and what your data passes through them.
- Whether any tool vendor uses your prompts or data to train models; the questions to put to vendors are in how to find out if a visibility vendor trains on customer prompts.
- A data processing agreement where personal data is involved, and your standard security questionnaire.
- How confidential material, such as roadmap details, is kept out of prompts.
7. Commercials
- Pricing model: retainer, project or hybrid, with what each fee includes.
- Tool costs: included or passed through.
- Contract length, notice period and termination rights.
- A statement that no specific placements or rankings are guaranteed.
Budgeting context is covered in whether AEO should have its own budget line or come from the SEO budget.
8. Submission and timeline
State the response format, page limit, deadline, date for questions, and when you will decide.
What should you prepare before sending it?
A few internal steps make responses far more comparable:
- Agree the objective internally. Marketing, product marketing and sales should agree which segments and categories matter most, or agencies will each guess differently.
- Draft 15–20 example questions that buyers ask, drawn from sales calls and win/loss notes. Include them in the RFP so agencies respond to the same reality.
- Gather what you already know: any AI visibility checks done so far, review platform profiles, and recent changes to pricing or positioning.
- Name an internal owner who will answer agency questions, approve the question set and review reports.
- Decide the budget range. Sharing it avoids proposals that are either far too small to matter or far too large to approve.
How should responses be scored?
| Criterion | Suggested weight |
|---|---|
| Measurement method and evidence standards | 25% |
| Relevant case evidence and references | 20% |
| Strategy and first-90-day plan | 20% |
| Team experience and time allocation | 15% |
| Price and commercial terms | 10% |
| Security and data handling | 10% |
Adjust the weights to your situation, but keep measurement high. An agency that cannot measure honestly cannot show whether anything it does works.
Should you include a paid test?
Yes, for shortlisted agencies. Give each the same ten questions and one competitor, and ask for a short audit within two weeks. The audit should cover current answers across engines with sources, the main reasons for the result, and three recommended actions. Pay a modest fixed fee. The test shows how each agency works with real data far better than a proposal document does. Before comparing the audits, check them against the standards in what an agency should check before sending a client an AI visibility report.
Which answers are red flags?
- Guaranteed placements in ChatGPT or other assistants.
- Claims of secret ranking factors or special relationships with AI companies.
- Reports with screenshots but no stored answers, sources or raw data.
- Case studies with no method or time frame, or only branded questions.
- Plans centred on publishing large volumes of AI-written pages.
- Undisclosed posting in community forums or review sites on your behalf.
- No mention of how run-to-run variation is handled.
What might an illustrative RFP process look like?
Consider an illustrative project management software company selling to mid-sized agencies. It sends the RFP to six agencies and receives five responses. Two are cut for guaranteeing placements. Three are invited to the paid test with the same ten questions. One returns screenshots of single runs. Another returns a thorough audit, but builds its main recommendation on branded questions. The third runs each question three times in five engines and shares the raw data. It shows that the company is absent from answers about agencies, and traces that to review platform categories and a missing comparison page. The company chooses the third agency, on a six-month contract with a review at month three against the agreed question set.
What do people ask about AEO agency RFPs?
How long should the RFP be?
Six to ten pages is usually enough. Longer RFPs deter good agencies without adding useful detail.
Should the RFP include SEO as well as AEO?
If one team or agency handles both, yes. Many B2B software companies combine them, since search rankings still feed many AI answers.
How many agencies should receive it?
Four to six, leading to a shortlist of two or three for the paid test.
What contract length is reasonable?
Consider six months with a review point. It gives time for work to show in answers without a long commitment.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources