How to choose an AEO agency: the questions to ask, the red flags, and what a managed AEO service should deliver
Key point. Choose an AEO agency on its method, not its label or its place on a list. A good one names the engines and buyer questions it will measure, takes a repeated baseline before it changes anything, keeps some questions untouched as a comparison, and reports results per engine as ranges. Walk away from guarantees.
AEO agencies, GEO agencies and AI SEO agencies sell the same kind of work under different names: getting your company read, cited, described correctly and recommended in the answers ChatGPT, Google's AI Overviews, Perplexity, Copilot and Claude write for your buyers. This guide is for founders and marketing leads at SaaS and tech companies, fintech, and other growing businesses who are deciding whether to hire one and how to pick. We sell managed AEO services ourselves, so we have kept this guide to criteria you can apply to anyone, us included, and we name no other firm.
Key takeaways
- The label tells you little. AEO, GEO, AI SEO and LLMO describe mostly the same work; ask which engines, which questions and which outcome are measured.
- Lists are a weak guide. In our measurement of AI answers to a "best firms for AEO" question, most cited sources were "best AEO agencies" lists, several written by agencies ranking themselves.
- A good agency measures before it acts. Expect a written question list, a baseline asked several times on more than one day, and untouched comparison questions.
- Expect results per engine, as ranges. A single blended "AI visibility score" or a before-and-after screenshot cannot show the work caused a change.
- Guarantees are a red flag. Google says no third party can guarantee performance, and no one controls what an AI engine writes.
- Start with a short pilot on one stage, with the success threshold agreed before the work starts.
What does an AEO agency do?
An AEO agency, or a managed AEO service, does the answer engine optimization work for you: it finds where your company drops out of AI answers and fixes that stage, then measures whether the fix moved anything. "Managed" means the team does the work with you, rather than handing you a dashboard or a report to act on alone.
The work follows the path an answer takes. It checks that each engine's search crawler can fetch your pages, that your pages are retrieved for the questions buyers ask, that engines trust your site and the independent sites that describe you, that they describe your company as it is today, and that they recommend you when a buyer asks for options. Our definition of AEO and the AI search visibility funnel describe each stage.
It is useful to separate three things that are often sold together:
| Option | What you get | Fits when |
|---|---|---|
| Managed AEO service (agency) | A team that diagnoses, does the page, site and third-party work, and measures the result | You have no one in-house with time for it, or you want it done to a research standard |
| AI visibility tool or platform | Software that asks engines your questions and reports what they say | You have a team that can act on the findings |
| In-house with an SEO lead | Your own team, using Search Console, Bing Webmaster Tools and a question list | You have SEO capacity and a small number of engines to cover |
Many teams combine them: a platform for monitoring, and a managed service for the stages that need hands-on work.
Do you need an AEO agency at all?
You need help when your buyers use AI answers to build a shortlist and you can see you are missing from it, and you do not have the time or the measurement skills in-house. You may not need one if engines already describe and recommend you correctly for your buyers' questions.
Run this check before you call anyone. Ask ChatGPT and one other engine five questions a buyer would ask, three times each: what your company does, which category you are in, whether you are credible, how you compare with an alternative, and "best [category] for [use case]". If you are named, described correctly and listed in most answers, an agency has little to fix. If you are missing, described out of date, or never recommended, you have found the stage to fix, and a better brief for any agency you speak to.
Google's hiring advice for SEO makes a point that carries over: knowing the basics yourself helps you spot a provider who wants to use a technique that is not recommended. Our guide to how to do prompt research shows how to build the question list.
How should you read "best AEO agency" lists?
Read them as marketing, and check who wrote each one. AI engines answer "which agency should I use?" largely from these lists, so the lists shape the shortlist your own buyers see, and many are written by the agencies they rank.
We measured this when we asked two AI engines about our own category. For the question about which firms are best for AEO for a SaaS company, "most sources cited for the category question were 'best AEO agencies' list articles, several by agencies ranking themselves." That measurement covers the OpenAI API with web search and the Perplexity API, 12 questions asked three times each on one day in October 2026, not the consumer apps.
The same pattern shows up in lists of AI visibility tools: "When we read the lists that AI engines cited for these buying questions, seven of the eight were published by companies that sell an AI visibility or SEO tool (95% interval 53% to 98%), and all seven ranked their own product first (95% interval 65% to 100%)." And: "None of the eight lists we read published such a test (0 of 8; 95% interval 0% to 32%)", meaning a published result of the tools run on the same questions. Our guide to what an AI visibility audit covers explains what a real diagnosis contains.
A list can still be a useful starting point for names. Do not read a position on it as evidence. Apply the questions below to every name instead.
What should a managed AEO service deliver?
A managed AEO service should deliver a measured diagnosis, a change plan tied to one stage at a time, the work itself, and a report that compares the questions worked on with the ones left alone, so you can tell a real change from normal movement. Ask for each of these by name, and ask to see an example.
| Deliverable | What good looks like |
|---|---|
| Buyer question list | 20 to 60 questions in your buyers' words, split into branded and non-branded, agreed with you and kept fixed during the test |
| Baseline | Every question asked at least three times per engine, on at least two days, before any change |
| Crawler and index check | Each engine's search crawler checked by name in robots.txt and at the firewall; Google's Search generative AI setting; Bing Webmaster Tools |
| Description audit | How each engine describes your company today, compared sentence by sentence with your current positioning |
| Source map | Which sites each engine cites for your questions, and which of them say something about you |
| Change plan | One stage at a time, with the prediction and the success threshold written down before the work starts |
| Comparison questions | Similar questions or pages left untouched and measured on the same days |
| The work | Page changes, factual consistency across your profiles, and outreach to the independent sites engines already cite, with each change logged by date |
| Monthly report | Per engine: how often you are cited, named, described correctly and recommended, each with the number of answers behind it and a range |
| Your data | The question list, raw answers and change log handed over, so the results outlive the engagement |
The last row matters more than it looks. If the raw answers stay inside the agency's tool, you cannot check its figures or carry your baseline to the next provider.
Which questions should you ask an AEO agency before you sign?
Ask these twelve, in writing, and compare the answers side by side. A good agency answers each one plainly and in numbers where numbers apply.
- Which engines will you measure, and why those? A good answer names them (for example ChatGPT, Google AI Overviews and AI Mode, Perplexity, Copilot, Claude, Gemini) and picks them from where your buyers are, not from what its tool covers.
- Which questions will you track, and who writes them? A good answer starts from your buyers' words, separates branded from non-branded questions, and fixes the list before the test.
- What exactly do you count? Citations, mentions, accurate descriptions and recommendations are four different things, and a good agency reports them separately per engine.
- How many times do you ask each question, and on how many days? AI answers change between identical asks, so a good answer is at least three asks per question per engine, repeated on another day.
- How do you tell a real gain from normal movement? A good answer describes a baseline range and a comparison group of untouched questions or pages.
- What would count as failure, and when? A good agency writes the threshold and the date before the work starts.
- Which part of the problem do you expect to fix first, and why? A good answer comes from a diagnosis of your company, not from a standard package.
- Which tactics do you use, and which do you not? Check them against Google's own guidance: for Google Search, Google says llms.txt, "chunking" content, rewriting just for AI and special schema markup are not needed.
- How do you work with the sites engines already cite about us? A good answer is about accurate, consistent facts on real profiles and publications, never paid or fake mentions.
- Will you share every change you make, with dates, and the raw answers? Google's SEO hiring guide asks the same of any provider; the change log is what lets you tie a result to the change made before it.
- What access do you need? Read access to Search Console and analytics is enough for a diagnosis; Google's guide advises granting read access only at the audit stage.
- Can you show a past result with its comparison group? A good answer shows the measured change, the number of answers behind it, and how the untouched questions moved over the same period.
What are the red flags when hiring an AEO agency?
The red flags are promises nobody can keep and measurements that cannot show cause. Any one of these is a reason to ask harder questions; two or more is a reason to keep looking.
- A guarantee. "We guarantee ChatGPT will recommend you" or "first place in AI Overviews". Google says third-party tools "can't guarantee performance", and no one controls what an AI engine writes.
- A special relationship. Claims of a partnership, an approval or a "priority" route into an AI engine's answers. Google says it "doesn't evaluate third-party services".
- One blended score. A single "AI visibility score" that averages engines together hides which engine moved and why.
- Screenshots as proof. A before-and-after screenshot of one answer cannot separate the work from normal movement.
- Tactics as the strategy. A plan whose core is llms.txt files, schema markup or rewriting pages "for AI", which Google says its Search does not need.
- Mass content as the plan. A proposal built on publishing dozens or hundreds of AI-written articles a month. Google's spam policies list "using generative AI tools or other similar tools to generate many pages without adding value for users" as scaled content abuse, and more pages do not fix the stage where you drop out.
- Fake mentions. Buying reviews, placing anonymous forum posts or seeding listicles. Google warns that seeking inauthentic mentions "isn't as helpful as it might seem", and the risk to your reputation is real.
- No comparison group. Reports that show only the questions the agency worked on.
- A secret method. An agency that will not explain what it is changing on your site, or why.
- A ranked list where the author is first. The agency's own "best AEO agencies" article with itself at the top is advertising, not evidence.
How do you know whether an AEO agency is working?
You know it is working when the questions or pages it changed moved by more than the ones it left alone, and by more than the range measured before it started. Anything less is a reading, not a result.
This matters because AI answers move on their own. "When we checked the same 118 searches three times over two days, only 14 of them returned the exact same set of cited websites each time." And: "Whether an AI Overview appears at all also moves. In the same three checks over two days, 22 of 150 searches showed an AI Overview on some checks and not on others (15%; 95% interval 10% to 21%), so a KPI that counts AI Overview appearances can change with nothing changed on your side."
A comparison group is what separates the work from that movement. In a simulation built on the structure of our own Google AI Overview data: "an analysis that did not know which searches had been worked on never detected an injected improvement, even a 20-point one (0 of 500 simulations at each size; 95% interval 0% to 0.8%). Told which searches had been worked on, the same analysis detected a 5-point improvement in 56% of simulations (95% interval 52% to 60%) and a 20-point one in all 500."
Most advice skips this step. When we read the KPI guides Perplexity cited for questions about AI search ROI: "The eight guides we read list sensible KPIs, but none of them says how to tell a real gain from that movement, and none recommends comparing the questions you worked on with questions you left untouched (0 of 8; 95% interval 0% to 32%)."
Two first-party reports give you an independent check on any agency's Google and Microsoft figures. Google Search Console's Generative AI performance report counts how often links to your site were shown in AI Overviews and AI Mode, and Bing Webmaster Tools reports citations in Microsoft's AI experiences. Open both yourself and keep your own baseline.
How should an AEO engagement be structured?
Structure it as a short pilot on one stage with a readout agreed in advance, then decide whether to continue. That protects you from paying for months of activity that cannot be judged.
- Weeks 1 and 2: diagnosis. The question list, the crawler and index check, and a baseline of every question asked several times on at least two days, per engine.
- Weeks 3 and 4: the plan. The first failing stage, the prediction, the success threshold and the comparison questions, written down.
- Weeks 5 to 10: the work. One set of changes, each logged with its date.
- Weeks 11 to 13: the readout. Both groups measured again on the same schedule and read as a range; then continue, change course or stop.
Size the readout before week 1. With a few dozen questions, only large changes will be visible, and a good agency says so up front.
What changes for a SaaS company or a startup?
For a SaaS company, fintech or funded startup, the questions that matter most are often comparison and diligence questions, not only "best X for Y". A buyer may ask an assistant whether you are credible before a demo, and an investor may ask before a round. So ask an agency how it measures what engines say about your credibility, your current product and your category, and how it handles a recent rebrand or a move upmarket, where engines often describe the older company.
For any growing business, the second difference is scale. A small team needs an agency that works through one stage at a time and hands over what it learns, not one that needs a large content budget before anything can be measured.
How does Elementera's managed AEO service fit these criteria?
We built our managed AEO service around the method on this page, so you can hold us to it. We work through the stages one at a time with your team, starting from your buyers' questions, and we follow four rules on every engagement and every finding we publish:
- We ask each question more than once. In our research measurements every question is asked at least three times, and repeated on another day before we believe a change.
- We never pool engines. ChatGPT, Google AI Overviews, Perplexity, Claude and Gemini read different parts of the web, so a result measured on one engine is reported for that engine only.
- We publish a finding only with its scope. Every finding we quote names the engine, the kind of question and the sample it was measured on.
- We record which model answered. Every measurement records the model we asked for and the model that actually answered.
Our founders have published machine learning research since 2015, including early work on generative audio and speech such as SampleRNN and Char2Wav. Lenz, our AI visibility platform, gives you a free first look at how ChatGPT and Gemini answer your buyers' questions, so you can see the starting point before you talk to anyone. Ask us the twelve questions above; we would rather be chosen on the answers.
Frequently asked questions
Is an AEO agency worth it?
It is worth it when your buyers build shortlists from AI answers, you are missing or misdescribed in them, and you lack the time or measurement skills in-house. Measure your starting point first; if engines already describe and recommend you correctly, spend the budget elsewhere.
What is the difference between an AEO agency and an SEO agency?
An SEO agency works to rank your pages in search results. An AEO agency works on what happens after retrieval: whether AI engines read your pages, describe you correctly and recommend you. The foundations overlap, and many SEO agencies now offer AEO, so ask both kinds the same questions.
Can my current SEO agency do AEO?
Often it can do the foundation, crawlable and indexed pages that answer the question. Ask whether it can also measure each engine separately with repeated questions, audit how engines describe you, and work on the third-party sites engines cite. If it reports only rankings, it is not measuring AEO.
Should I hire a GEO agency or an AEO agency?
The labels describe the same work. Choose on method: which engines are measured, which outcomes are counted, and how a change is proven.
Which companies offer managed AEO services for growing businesses?
Specialist AEO firms such as Elementera, and a growing number of SEO and content agencies that have added AEO. Use the twelve questions above to compare them; a provider's place on a "best agencies" list tells you little about its method.
What is the best AEO agency for a B2B SaaS startup?
The one that measures the questions your buyers ask, on the engines they use, and can show a past result against a comparison group. For a startup, also look for a short pilot, a clear handover of data, and experience with comparison and diligence questions.
Is publishing hundreds of AI articles a good AEO strategy?
No. Volume does not fix the stage where you drop out of AI answers, and Google's spam policies treat generating many pages with AI tools "without adding value for users" as scaled content abuse. A few pages that answer your buyers' questions plainly, with the same facts stated on the independent sites engines read, are the better use of the budget.
How long does an AEO agency take to show results?
Plan for a readout after about three months on one stage. Some fixes show sooner: OpenAI says a robots.txt change reaches its search results in about 24 hours. Fixes that depend on other websites changing what they say about you can take months.
Can an AEO agency guarantee results?
No. No one controls what an AI engine writes, and Google says third-party tools cannot guarantee performance. What an agency can commit to is a method, a written success threshold and an honest readout.
Work with us
If you want the AEO work done for you, our managed AEO services follow the method on this page: Book a call. If you would rather see where you stand first, Lenz gives you a free analysis of how ChatGPT and Gemini answer your buyers' questions: Get your free AI visibility report.
How this page was made
Dr. Mehrdad Mehrbod wrote this page with the help of an AI writing assistant (Claude, made by Anthropic), working from Google's published guidance on choosing search providers and from our own measurements. Before writing, we read the pages that rank on Google for "aeo agency" and "ai seo agency" and the sources AI engines cited for "Which companies offer managed AEO services for growing businesses?" on October 13, 2026. We name no other provider and rank no one. Findings of ours appear only as we have published them. Dr. Maziar Gomrokchi reviewed the page, and Dr. Mehrdad Mehrbod approved it for publication.
Sources
- Google Search Central, "Do you need an SEO?", last updated June 5, 2026: developers.google.com
- Google Search Central, "Google Search's guidance on using third-party SEO tools, services, and advice", last updated June 5, 2026: developers.google.com
- Google Search Central, "Spam policies for Google web search" (scaled content abuse), last updated August 28, 2026, verified October 20, 2026: developers.google.com
- Google Search Central, "Optimizing your website for generative AI features on Google Search", last updated July 10, 2026: developers.google.com
- Google Search Console Help, "Generative AI performance report (Search)": support.google.com
- OpenAI, "Overview of OpenAI Crawlers": developers.openai.com
- Elementera, measurement of the OpenAI API with web search and the Perplexity Sonar API: 12 questions about our company and category, three times each, October 5 and 6, 2026.
- Elementera, reading of the 8 AI visibility tool lists and 8 KPI guides that AI engines cited for buying and ROI questions, October 2026.
- Elementera, measurements of Google AI Overviews on 150 non-branded US searches, three checks over two days, July 2026, and a simulation built on the structure of that data.
Known professionally as Dr. Med with 12+ years of experience implementing AI products. Strategic advisor to multiple successful founders including a unicorn, US patent holder. Ex-SOTI, UC Berkeley PhD.
AI/ML scientist with 20+ years of research & industry experience. Ex-Google DeepMind, MIT, Harvard, Mila. McGill-Mila PhD.
Related articles
- Why Choose Answer Engine Optimization (AEO) for AI Search Optimization Over Traditional SEO?
- AI search glossary: 55 terms, defined precisely
- Is good AEO just good SEO? What our measurements show
- What an AI Visibility Audit Actually Covers for SMBs
- Inside Google’s First Year of AI Mode: What 1 Billion Users Reveal About How Americans Search Now
Latest articles
- Elementera receives angel investment to grow Lenz and the team behind it
- Dental Practice Software in 2026: A Practical Guide to PMS Platforms and AI Tools (US and Canada)
- Google Preferred Sources and AEO: what it is and how it affects your AI visibility
- What You Might Have Missed From Google I/O 2026 (And Why It Matters for AEO and AI Search)
- What Is llms.txt and How to Implement It for AI Bots (2026 Guide)