Most buyers asking how to choose an SEO agency for AEO visibility already know classic SEO. They can spot a thin audit. They have been burned by link farms. What they struggle with is the new pitch: every agency now claims ChatGPT, Perplexity, and AI Overviews expertise, and most of those claims collapse the moment you ask for dated citation proof.
This guide is a commercial investigation checklist, not a soft overview. You will leave with a scoring rubric, questions that force real answers, red flags that matter for AEO specifically, and a clear call on when to hire versus build in-house. If you only need a classic SEO partner, use our separate guide on choosing an SEO services company. Here the bar is different: can they move how AI systems cite you, and can they prove it?
If you are still aligning the vocabulary, start with what is AEO and what is generative engine optimization. Then come back here when you are ready to shortlist partners for AEO services.
What "good" looks like for AEO, not SEO
Traditional SEO evaluation leans on rankings, organic sessions, and keyword coverage. Those still matter. For AEO they are incomplete. Answer engines do not hand you a stable blue-link report. They synthesize. Your brand either gets named, linked, or skipped inside that synthesis.
A credible AEO agency should treat citation rate and share of voice as primary KPIs, not vanity slides. Citation rate is how often your brand appears in answers for a locked prompt set. Share of voice is your citations versus named competitors on that same set. If a proposal only promises "AI-optimized content" and "schema upgrades" with no measurement plan, you are buying activity.
- Multi-engine sampling across ChatGPT, Perplexity, Gemini, and Google AI Overviews - not one chatbot demo
- A stable prompt set tied to buyer questions, not cherry-picked vanity queries
- Before/after citation evidence with dates, not screenshots without context
- Entity and authority work that goes beyond on-page FAQs
- Reporting that connects visibility movement to pipeline or qualified demand when the deal size justifies it
If the agency cannot show how it measures citation rate and share of voice on a real client, stop the evaluation. Strategy decks without a measurement stack are marketing, not AEO delivery.
The AEO agency scorecard
Score each shortlisted agency from 1 to 5 on the criteria below. Do not average blindly. A 2 on measurement or citation proof should outweigh a polished sales process. Weight the first three rows more heavily than commercials if your goal is AI visibility, not a cheaper content mill.
Practical AEO agency scorecard. Score 1 to 5 per row. Treat measurement, citation proof, and methodology as veto criteria - a low score there is usually enough to drop the vendor.
| Criterion | What a 5 looks like | How to verify in discovery |
|---|---|---|
| Measurement stack | Weekly or continuous sampling across 4+ answer surfaces with exportable prompt/response pairs | Ask for a live dashboard walkthrough on a current client, not a mockup |
| Citation proof | Dated before/after citation rate and share of voice on commercial prompts | Request two anonymized client examples with dates and prompt lists |
| Retrieval methodology | Clear explanation of how content, entities, and third-party sources get into answers | Ask them to map one of your money prompts to specific workstreams |
| Entity and authority | Plan for brand consistency, corroborating sources, and off-site mentions - not FAQ schema alone | Ask what they would fix if on-page content was already strong |
| Technical depth | Schema, crawlability for AI bots, answer-shaped pages, and QA for releases | Have them critique one of your key templates live |
| Commercial honesty | Month-to-month or fair exit terms, no ranking guarantees, clear scope boundaries | Read the MSA for success definitions and cancellation clauses |
Use a simple rule: shortlist only agencies that clear a 4 or 5 on measurement and citation proof. Everything else is negotiable. Those two are not.
Questions that separate specialists from buzzword SEO
Sales calls reward confident storytelling. Force specificity. Compare answers across vendors on the same questions, then judge who can name engines, metrics, and workstreams without collapsing into "we also do AI SEO."
- Show me citation rate and share of voice for a live client on ChatGPT and Perplexity for the last 90 days.
- Which buyer prompts would you baseline in week one for our category, and why those prompts?
- What would you refuse to do if we asked for page-one Google rankings in 30 days as the AEO KPI?
- If our content is already strong, what off-site or entity work would you prioritize next?
- How do you handle answer volatility when the same prompt returns different citations week to week?
- Which part of AEO should stay with our in-house team, and which part should you own?
Weak answers sound like recycled SEO: more blogs, more backlinks, more schema, with AI mentioned as flavor. Strong answers name surfaces, prompt sets, corroborating sources, and a measurement cadence. They also admit what they cannot control.
When you want a partner that already runs this stack end to end, compare specialist AI SEO agency models against generalist retainers. AEO is usually the wrong place to buy the cheapest full-service bundle.
Red flags that matter more for AEO than for classic SEO
Some classic SEO red flags still apply. Guarantees, opaque link tactics, and one-size packages remain bad. AEO adds a few sharper ones because the category is new enough for marketing to outrun delivery.
AEO-specific red flags. One alone is a warning. Three or more usually means the vendor is selling SEO activity with AI vocabulary.
| Red flag | Why it matters | What to do |
|---|---|---|
| SEO firm with AI buzzwords only | Homepage leads with rankings and traffic; AEO is a slide, not a practice | Ask for citation KPIs. If none exist, treat them as an SEO vendor |
| Guaranteed AI citations | No agency controls model retrieval or competitor movement | Walk away. Process guarantees are fine. Outcome guarantees are not |
| Under-30-day ranking or citation promises | Meaningful citation movement on competitive prompts rarely fits a vanity sprint | Demand a 90-day measurement plan with baselines instead |
| Traffic-only reporting | Organic sessions can rise while AI answers still ignore you | Require citation rate and share of voice in the monthly report |
| Schema as the whole strategy | Markup helps machines parse pages. It does not create third-party corroboration | Ask what they do for sources AI already trusts in your category |
"Guaranteed citations in 30 days" is the AEO version of guaranteed #1 rankings. Same sales psychology, same lack of control. Demand accountable process, not fictional certainty.
Proof package you should demand before signing
Do not accept "trust us, we have done this." Ask for a proof package that a non-technical executive can review in 20 minutes. If the agency cannot assemble it, they are not ready to own AEO outcomes.
- Baseline method: how they will lock prompts, engines, and sampling frequency for your brand
- Two dated case examples showing citation rate or share of voice movement
- Sample monthly report with AI visibility metrics, not only Search Console charts
- Named owners for content, technical, and authority workstreams
- Written scope of what is out of scope - especially paid media and product roadmap decisions
If the site has unresolved crawl, indexation, or template debt, start with SEO audit services before an AEO retainer. Answer engines still need a clean, extractable site. Paying for citation strategy on a broken foundation wastes budget.
Hire vs build in-house
Not every company needs an AEO agency. Some need a tool and a senior owner. Some need a specialist partner because the opportunity cost of slow internal learning is higher than a retainer. Be honest about bandwidth and deal size.
When to hire an AEO agency versus building capability in-house. Match the operating model to bandwidth and how much AI-referred demand already matters to revenue.
| Situation | Better path | Why |
|---|---|---|
| Strong SEO team, low AI referral share today, time to experiment | In-house + measurement tools | You can learn the craft while the stakes are still moderate |
| Marketing bandwidth is maxed, AI answers already influence deals | Specialist AEO agency | You need execution velocity, not another dashboard for a busy team |
| Enterprise deal sizes, multi-product entity complexity | Agency or hybrid with senior in-house owner | Entity and governance work needs depth and cross-team coordination |
| Small site, local or simple category, limited budget | Focused packages or audit-first scope | A full AEO retainer may be overkill before fundamentals are solid |
Smaller teams that still need structured help can look at SEO packages for small business as a starting operating model, then layer AEO measurement once the site can support answer-shaped content. Authority gaps often still need intentional link building services because AI systems repeatedly lean on sources the open web already trusts.
A 30-day evaluation plan
Treat vendor selection like a short project, not a vibe check. Two weeks of discovery beats six months of the wrong retainer.
- Week 1: Lock your own baseline - 25 to 50 buyer prompts, sample ChatGPT, Perplexity, Gemini, and AI Overviews yourself or with a tool.
- Week 2: Shortlist three agencies. Send the same RFP questions and scorecard.
- Week 3: Run live demos of measurement stacks and review proof packages.
- Week 4: Score vendors, check references, negotiate scope and exit terms, then pick one.
Your own baseline is the underrated step. When agencies pitch diagnosis, you can grade accuracy against a known truth. Buyers who skip that step get sold the most confident story, not the best operator.
The bottom line
Choose the AEO agency that can measure citation rate and share of voice, show dated proof, and explain retrieval work beyond blogging plus schema. Fire the ones who guarantee citations, promise under-30-day miracles, or report only classic SEO metrics with an AI slide stapled on.
When you are ready to evaluate a specialist partner, start with how we approach AEO services and compare that against any generalist proposal. The right hire will feel slightly uncomfortable in discovery - because they ask harder questions than the vendors who only want to close.
