Does AEO / GEO Work? It Depends On Who You Ask.
You can't go a day without hearing the word AI.
It's the hottest business topic out there right now. And just like 1999, when everyone was launching a .com, everyone today is putting out a .ai app, tool, service, or offering.
In marketing, no topic is hotter than being found by AI search. Depending on who you ask, it goes by several different names:
AEO (Answer Engine Optimization)
GEO (Generative Engine Optimization)
LLMO (Large Language Model Optimization)
AIO (AI Optimization to some vendors, Google's AI Overviews to others)
AI SEO
AI visibility
AI share of voice
They are all sold with dashboards: tracking dozens of queries and their fan-out queries (the related searches AI runs behind the scenes), counting your mentions, your citations, and what AI thinks of your brand.
ChatGPT launched in November 2022. One year later, researchers at Princeton coined GEO.[1] And... BAM! An entire new industry was born. Not even three years old and it's got more than a half dozen names. That's not a mature industry. That's an industry trying to figure itself out.
And while there are differences between them, that's not the important question.
The question you should be asking is:
Does any of this AI search stuff even work?
Well, the answer depends on who you ask.
The marketing industry will tell you AEO works.
Ask any marketing agency, any of the big marketing platforms (HubSpot, Semrush, Ahrefs), or any of the hundreds of AI visibility startups behind them, and you'll get a passionate yes: AEO works!
I can't go a week without getting hit up on LinkedIn or getting an email from a new AI search company I've never heard of. Today, as I was writing this article, I ran into one that was one week old. And every AI search company has the data to prove it works.
Semrush published a case study on itself: after two months of targeted content updates, its AI share of voice rose from 13% to 32% across its tracked prompt set. A 146% lift, measured in its own tool.[2]
Marketers love data. I know, I'm one of them. So you can bet all of these companies have a bunch of KPIs to prove AI citations work:[3]
Share of Model: how often your brand appears in AI answers for your category
Share of voice: your mentions as a percentage of the total
Citation rate: how often AI cites your content as a source
Sentiment score: how positively AI describes you
Competitive rank: how your visibility stacks up against competitors
And all of these metrics get wrapped into a pretty report every month to show progress.
Here's the thing: under the colorful dashboards, numbers, and graphs is a set of beliefs.
AI changed where buyers search, not how they behave. The eyeballs moved from the search results page to the AI answer.
Buyers are using AI as a search engine. A single turn with a long-tail conversational prompt, and the goal is to show up in the answer as a citation or a mention.
The old inbound playbook is still a reliable source of leads. Target long-tail keywords, write the content, get the website visitors, capture the email, nurture the list. It worked for Google, just point it at AI now.
All three add up to one bet: new medium, same playbook.
Here is how you see it in their proposals and pitch decks:
Track a fixed number of prompts, like "best CRM for small business" or "top IT services companies in Orange County." Entry plans typically start you with a small set of tracked prompts, and the bigger published tiers run to a few hundred.
Run the prompts against multiple AI models, like ChatGPT, Gemini, Perplexity, and Claude, on a daily schedule.
Record and track which brands get mentioned and cited.
Chart the counts and wrap them into the monthly report.
The price of the tracking tools behind these packages? Entry plans start around $29 a month, with top published tiers running several hundred dollars, and custom enterprise pricing above that. The agency's fee for running the program sits on top.[4]
And look, I get it. I used to own a HubSpot agency. To be honest, if I were still an agency owner, I'd probably be selling an AEO package too. Because it's an easy sell. Metrics that can be measured, tracked, and reported on every month. And the actual tactical work isn't too different than traditional SEO.[5]
The management consultants will tell you something different.
Ask the strategy consultants who advise the Fortune 500 (Forrester, Gartner, McKinsey) and you'll hear a different story. These firms get paid to figure out what changed, so their clients can make informed decisions about where to spend their marketing budget.
And here is what they found.
2024: Forrester reported that 89% of B2B buyers had already adopted generative AI, and named it one of the top sources of self-guided information in every phase of their buying process.[6] Every phase. Not just finding vendors, but evaluating the differences between them and justifying the purchase commitment.
2025: Forrester wrote that buyer questions had moved beyond simple keyword matches. Buyers now expect AI to understand their intent and deliver answers that are contextually relevant, credible, and actionable, guiding them through complex decisions.[7]
2026: Forrester surveyed nearly 18,000 business buyers, and found that twice as many named generative AI or conversational search a more meaningful source of purchase information than any other; more than your website, your product experts, and your sales reps.[8] Note the scope of what Forrester found: not just finding vendors. Purchase information, in every phase of the decision.
Let me translate all that consultant speak: AI wasn't just a technology change; it changed the way buyers behave.
Your buyers aren't searching for vendors. They're asking AI to recommend one. And it's easy to see why: outsourcing vendor research and comparison to AI saves time. Research that used to take days now gets done in hours.
When buyers outsource vendor research to AI, the AI needs context to deliver relevant answers. So buyers start by telling their AI research assistant three things:
Who they are
What problem they're trying to solve
What they're looking for in a vendor
Once the AI has that context, it goes out and does the heavy lifting: finding the information, comparing the vendors, narrowing the options, and building the case the buyer takes to their leadership team.
This is not a one-turn informational search prompt to find a vendor. This is a multi-turn conversation with their AI research assistant.
Notice what the consultants lead with. Not AEO, not GEO, not share of voice. They lead with buyer behavior, and any advice about showing up in AI search comes second, downstream of the behavior change.
And so with the management consultants, it's not about if AEO works or not. They're saying buyers are using AI as a research assistant more than a search engine. And once you understand that difference, it changes the whole debate.
So why is the marketing industry selling you AI search?
Here's the thing: marketers see the same trend the management consultants do. They see buyers googling less. They see buyers turning to AI for vendor recommendations.
So why is the entire marketing industry talking so much about AI search, visibility, citations, and mentions instead of recommendations?
While it's true that AEO is an easy sell, the root cause is a bit more complex than that.
The root cause is the business model. Agencies were created and then scaled on the search-era economy: monthly retainers, recurring deliverables, content production, monthly reporting. Recurring revenue from recurring work, done by teams built up over years.
Now, if you view AI as a technology change, then a lot from the standard SEO playbook survives. Same skills, same team, same retainer. Just point the playbook at a new medium.
But if AI is a buyer behavior change, the revenue model no longer fits as well. Adapting means a structural shift in how agencies make money, how they pay their teams, and what work they do.
With a structural shift, adapting means more than retraining and incrementally improving the team.
An agency owner is faced with an uncomfortable question: what does this mean for the team of people who have worked beside them for years?
At the enterprise level, the structural shift is already underway. The global ad market grew 8.6% in 2025 while revenue at the world's biggest marketing agencies fell, and those same agencies have cut tens of thousands of roles.[9] And what happens at the enterprise level is a leading indicator of what eventually reaches the mid-market.
Few businesses volunteer for that kind of change. And fewer owners want to.
Harvard professor Clayton Christensen had a name for this: the innovator's dilemma.[10] Incumbents can see a disruption clearly and still struggle to respond, because responding means acting against everything, and everyone, that built the business.
For years, we marketers trained CMOs, CEOs, and entire marketing teams with colorful dashboards and monthly reports full of graphs, trend lines, and KPIs like ROAS, conversion rates, and MQLs. And they weren't wrong to trust them. That model worked in the inbound era: eyeballs, traffic, clicks, all of it measurable, all of it real, and it generated leads and revenue. Thousands of marketing agencies were spawned from inbound, including mine. SEO tools like Semrush became the standard. AEO feels safe because it's familiar. It looks like the reports everyone already knows how to read.
And to be fair, the AI visibility metrics do track something real. If what you're chasing is being found, then citations and mentions are worth watching. Just remember that AI search is one way of being found. There are dozens of others.
So it's not that marketers or agencies are lying about AEO. It's one legitimate way to be visible to your buyers. Just understand that agencies are reluctant to change what they sell, held in place by their business model and by loyalty to the people who built it with them.
So will agencies continue to sell AI search? Christensen answered that one too. Studying industry after industry, he found that leading firms were "held captive by their customers."[10] In other words, as long as customers kept buying the old thing, the incumbents couldn't justify changing. And by the time the buying stopped, it was usually too late.
Which means the marketing industry will keep selling AI search until enough SMB owners realize what the management consultants are telling their enterprise-level clients: AI has changed buyer behavior.
The change won't start with the agencies. It starts with the businesses that buy from them.
And it starts with the question hiding inside the consultants' findings: if buyers are using AI to compare, narrow, and recommend vendors, what decides who makes the shortlist?
The buyer decides who makes the shortlist, not the AI.
Being recommended by AI is not a technology challenge. There's no secret algorithm to crack. You don't need to adapt to every model update. That's because being recommended by AI is a business problem: matching the buyer's requirements against your capabilities.
Remember, the AI research assistant works from the context the buyer gave it: who they are, what problem they're trying to solve, what they're looking for in a vendor. AI does its analysis based on these buyer requirements.
When AI does its analysis, it does so on behalf of the buyer. So like a buyer, it asks the same six questions buyers have always asked when evaluating a vendor. You already know these questions. You've asked them yourself every time you hired a lawyer, an accountant, or an agency:
Do you solve my problem?
How do you do the work?
Do you work with companies like mine?
Can you prove your results?
Why should I trust you?
What does the engagement look like?
And this is why being recommended by AI is a business problem. These are the same questions a duke asked when hiring a master mason to build his cathedral eight hundred years ago. These qualification questions survived the printing press, the telephone, and twenty years of Google. They'll survive ChatGPT. The questions didn't change. Just who's doing the work to answer them.
So when the AI research assistant goes about its analysis, it draws from three main sources of data:
Training data: what the AI already knows
The buyer's context: the conversation, the requirements, the files they share
The web: what it can read at research time
Here's what matters about these three sources: control.
Training data: most small and mid-sized companies barely show up in it. And you don't control it: you can't call OpenAI or Anthropic and tell them to add you.
The buyer's context: their prompts, their requirements, their documents. You don't control that either.
Your website: the one source you fully control. External sources matter, but mostly for one question, why should I trust you. Your website is the place that can best answer all six.
And here's the important thing to know about adapting to the AI research assistant: missing data.
Before AI, when a human was researching a company on their website, if information was missing or unclear, the buyer could always contact the vendor to fill in the gaps ("Got a question? Contact us"). In fact, some companies deliberately kept their content thin so a buyer was forced to call the sales team.
When AI does the research and hits a data gap or conflicting information, it doesn't initiate a phone call, write an email, or start a chat. It just uses whatever information it can find. A missing answer could be a silent disqualification.
And what's worse, you'll never know you were disqualified, because the conversation between the buyer and AI is private. You just never hear about the deal that went to someone else.
The shortlist doesn't go to whoever published the most content. It goes to whoever answered the buyer's questions completely and correctly on their website.
So, does AEO work?
Sure, if what you're chasing is citations and mentions. If you care about the marketing metrics we've told you to watch for the last twenty years, the dashboards will deliver them. And is AEO a legitimate way for buyers to become aware of you? Yes. It's just one of many ways.
But as AI-generated content floods every channel, and it gets harder to tell what's real, human interaction is getting more valuable, not less: tradeshows, networking, referrals, recommendations. Think about it: would you rather ask someone you know and trust for a recommendation, or go out and search for a vendor yourself? For most small and mid-sized businesses, the referral still wins.
Now, if what you care about is revenue instead of visibility, you're better off spending your time and resources first on the foundation: making sure your website answers the buyer's questions completely and correctly.
And that doesn't mean "create more content" or "optimize for AI." That's treating a business problem like a technology challenge. It means making your website a database for AI by answering the six buyer questions on your foundational pages: a task with a finish line.
The people who can answer those questions are the people who interact with your prospects and clients every day: your sales team and the people who deliver the work. They already know the answers. But their superpower is probably not writing. For that, bring in an experienced content writer, in-house, freelance, or agency, to interview them, draw out that knowledge, and put pen to paper.
Once that foundation is solid, every acquisition channel works harder. It doesn't matter which one: referrals, conferences, tradeshows, cold calls, even AI search. There are many ways for a buyer to find you, but increasingly there's one practical way they evaluate you. All discovery channels lead to the AI research assistant, and the research assistant leans on your website data. Fix the foundation and you get a fair shake: evaluated on your merits, with your own answers.
And how do you measure that? The metrics that matter live at the bottom of the funnel, not the top. Less "how did they hear about us," because more and more, the honest answer is a conversation you'll never see. More: what information convinced this buyer to pick up the phone, and which buyer question did we answer well enough to make their shortlist.
Because here's the thing: if AI mentions you but can't tell the buyer why you're a good fit, being visible doesn't convert into being recommended.
So the real question isn't whether AEO works.
The real question is this:
Where do you want to spend your marketing budget?
Do you want to be a footnote?
Or
Do you want the recommendation?
Footnotes
[1] Aggarwal, P., Murahari, V., et al. "GEO: Generative Engine Optimization." arXiv, November 2023; published at KDD 2024. Researchers from Princeton University and IIT Delhi. The paper reports visibility gains of up to 40% from content optimizations, varying by domain. https://arxiv.org/abs/2311.09735
[2] Semrush Enterprise, "How Semrush Increased AI Share of Voice from 13% to 32% in Two Months." Self-published case study, measured in Semrush's own AI visibility tool. https://enterprise.semrush.com/case-studies/semrush/
[3] KPI definitions drawn from vendor documentation, including Similarweb's GEO guide (https://www.similarweb.com/blog/marketing/geo/what-is-geo/). Terminology and formulas vary by vendor; the industry has not settled on standard definitions.
[4] Prompt allowances and pricing from published vendor plan pages, including Otterly.ai, Profound, HubSpot AEO, and Semrush's AI toolkit, as of September 2026. Pricing changes frequently; verify with vendors.
[5] Google Search Central, "AI Features and Your Website." Google's own documentation states that optimizing for AI search largely follows standard SEO practice. https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
[6] Forrester, "B2B Buyer Adoption of Generative AI," November 2024 (https://www.forrester.com/report/b2b-buyer-adoption-of-generative-ai/RES181769); and John Buten, "The Future Of B2B Buying Will Come Slowly... And Then All At Once," Forrester blog, November 21, 2024 (https://www.forrester.com/blogs/the-future-of-b2b-buying-will-come-slowly-and-then-all-at-once/).
[7] Forrester, "From Keywords to Context: Impact and Opportunity for AI-Powered Search in B2B Marketing," October 2, 2025. https://www.forrester.com/blogs/from-keywords-to-context-impact-and-opportunity-for-ai-powered-search-in-b2b-marketing/
[8] Forrester, "Forrester 2026: The State of Business Buying," January 2026 press release, and "B2B Buyers Make Zero-Click Buying Number One," reporting Forrester's Buyers' Journey Survey 2025 (nearly 18,000 business buyers). https://www.forrester.com/press-newsroom/forrester-2026-the-state-of-business-buying/ and https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/
[9] Agency layoff and revenue reporting: Digiday, "As industry layoffs become the new normal, so does fear of AI's impact on adland's job market," November 2025 (https://digiday.com/marketing/as-industry-layoffs-become-the-new-normal-so-does-fear-of-ais-impact-on-adlands-job-market/); eMarketer, ad agency trends reporting, 2025 ad market growth of 8.6% against declining holding-company revenue (https://www.emarketer.com/content/ad-agency-trends-2026/).
[10] Christensen, Clayton M. The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business School Press, 1997. The quoted phrase "held captive by their customers" is from the book's analysis of the disk drive industry.