AEO and GEO ai search optimisation services: what they do and whether they work

TL;DR: AEO (Answer Engine Optimisation) and GEO (Generative Engine Optimisation) are the work of getting your brand into AI-generated answers instead of just ranking in blue links. Specialist services cost $2,000 to $25,000+ per month. The field is barely two years old, the evidence is thin, and most providers are repackaging old SEO. Here is how to tell the real work from the noise.

What are AEO and GEO, and why do they matter now?

AEO stands for Answer Engine Optimisation. GEO stands for Generative Engine Optimisation. Both describe the same job: getting your brand, product, or content cited when someone asks an AI assistant a question instead of typing a search into Google.

The split between the two labels is mostly marketing. In practice, AEO tends to mean optimising for assistants that pull structured answers, like ChatGPT, Perplexity, and Google's AI Overviews. GEO is the academic term, introduced in a 2023 Princeton and Georgia Tech paper that measured how edits to source documents changed how often generative models cited them [1]. The underlying problem is identical. Old-style SEO earns you a blue link. A growing share of people never click one. They read the AI's answer and move on.

Google's 2023 Search Quality Rater Guidelines acknowledge that AI-generated responses are changing how people use search results pages [2]. A 2024 study by Seer Interactive found Google's AI Overviews showed up in roughly 47 percent of sampled queries, though the number swung hard by category [6]. Navigational and transactional queries triggered fewer AI Overviews. Informational queries triggered the most.

If your business runs on informational search traffic, this shift hits you directly. If it runs on branded or transactional queries, the pressure is lower. Not zero, but lower.

See the broader picture at ai search and ai-powered search features.

What does an AEO or GEO service actually include?

This is where buyers get lost, because scope swings wildly from one provider to the next.

A credible AEO/GEO service should cover five things at minimum. One, a visibility audit: running your brand and category queries through ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews to document where you show up, where competitors show up, and how often. Two, content gap analysis: finding the questions the assistants answer without citing you, then mapping those to content you could credibly own. Three, structured data work: FAQ schema, How-To schema, Speakable schema that helps AI parsers read your content. Four, E-E-A-T signal work: author credentials, source citations, and factual density, all of which the GEO study linked to citation frequency [1]. Five, ongoing monitoring, because model weights shift with every update and a one-time fix decays.

Weaker providers take an existing SEO retainer, slap a new name on it, and add a monthly Perplexity check. That is not worthless. It is also not worth a premium. The tell is whether they can show you a before-and-after citation rate in a named AI engine on specific query clusters. If they cannot define the measurement, they cannot manage the work.

For what rigorous measurement looks like, the ai search visibility metrics kpis article is a good companion read.

How is AEO/GEO different from traditional SEO?

Traditional SEO optimises documents for a ranking algorithm that weighs hundreds of signals to order a list of links. The feedback loop is fast and measurable: rank position, click-through rate, organic traffic. You see movement in Google Search Console within days.

AEO and GEO optimise for retrieval-augmented generation systems and large language models that do not use the same signals Google's ranking algorithm does. Citations in ChatGPT's answers, for instance, often come from pages that rank well on Bing rather than Google, because ChatGPT's browsing uses Bing for web retrieval [3]. Perplexity runs its own crawler [7]. Google's AI Overviews lean hard on pages that already rank in organic Google search [8]. The work is not the same across targets.

The deeper gap is the feedback loop. There is no Google Search Console for AI citation. Nobody hands you an impression count for how often ChatGPT named your brand. You have to simulate it with prompt testing, which is expensive at scale and noisy by nature. Tools like ai-visibility-tool and ai seo tools are trying to close that measurement gap, with mixed success.

One more real difference. LLMs train on data with cutoff dates. Content published after a model's cutoff will not appear in non-retrieval answers until the model retrains or the system reaches for live web search. So for models without live retrieval, part of the optimisation window is already in the past. You are building a reputation the next model version will read.

Read more on the mechanics at ai seo and generative engine optimization.

What does the research say about what actually moves AI citation rates?

The GEO paper by Aggarwal et al. (2023) is the closest thing this field has to a peer-reviewed baseline [1]. The team tested nine content changes across 10,000 search queries and measured how each shifted citation frequency in AI answers. Adding authoritative quotes raised AI citations by up to 30 percent. Adding statistics lifted them by roughly 20 percent. Better prose lifted them by about 15 percent. Keyword stuffing did almost nothing, which confirms LLM retrieval does not behave like keyword-based ranking.

The study also found the winning tactic depends on the topic. Authoritative statistics matter most for science and technology queries. Quotes from named experts matter most for political and social questions.

A 2024 BrightEdge analysis found Google AI Overviews cited pages from the top 10 organic results about 74 percent of the time, which says traditional SEO authority still carries real weight for Google's AI product specifically [4]. Perplexity and ChatGPT with browsing wander further from Google's rank order, pulling from pages that rank strongly on Bing or that are well-structured but mid-ranked.

Nobody has good longitudinal data yet on whether AEO/GEO work holds up over six or twelve months. The honest read: the practice is under two years old commercially, and most agency case studies are one brand, one query set, one short window. That does not make the tactics wrong. It means the evidence bar is still low, and anyone claiming certainty is selling something.

For what Google's AI search rewards specifically, see google ai search.

Content change effect on AI citation frequency

| | | |---|---| | Added authoritative quotes | 30% | | Added statistics | 20% | | Improved prose fluency | 15% | | Keyword optimisation | 2% |

Source: Aggarwal et al. GEO study, Princeton/Georgia Tech, 2023

How much do AEO and GEO services cost?

Pricing is all over the map, which is what you expect from a new category with no settled benchmarks.

Project-based audits from credible boutiques run roughly $3,000 to $15,000, depending on how many AI engines they test, how many query clusters they analyse, and how deep the content recommendations go. Monthly retainers from specialist agencies run $2,000 to $25,000. The high end is mostly enterprise clients with multi-brand portfolios or regulated industries where every content change needs legal review.

Here is a rough breakdown of what you get at each price point.

| Monthly Budget | Typical Scope | |---|---| | Under $2,000 | Automated AI query monitoring only, no content strategy | | $2,000 - $5,000 | Visibility audit + content gap report + basic schema implementation | | $5,000 - $12,000 | Full audit + content production (4-8 pieces per month) + schema + monthly reporting | | $12,000 - $25,000 | Enterprise: multi-engine monitoring, PR and citation-building, author authority work, competitive tracking | | Over $25,000 | Custom, usually dedicated headcount or multi-agency coordination |

These ranges come from published pricing pages and RFP responses documented across the industry as of early 2025. Expect them to compress as more providers pile in.

Software tools (SaaS visibility monitors) run $300 to $3,000 per month depending on query volume and how many engines they cover. They do not replace strategy. They make the measurement part tractable. See brandrank.ai visibility insights analysis for a closer look at one such tool.

Which AI engines should you prioritise?

Google AI Overviews first, then Perplexity, then ChatGPT with browsing, then the rest. Here is the reasoning.

Google AI Overviews reach the biggest audience by a wide margin. Google's filings and announcements suggest they appear on a meaningful slice of the billions of daily queries, though Google has not published a precise number [2]. Because AI Overviews track closely to existing organic Google rankings [8], improving your Google SEO improves your AI Overview citation rate too. That makes it the most efficient bet for most businesses.

Perplexity has a smaller but unusually high-intent audience. Its users skew toward researchers, analysts, and technical buyers who go deep and follow citations. If you sell to knowledge workers or B2B buyers, Perplexity visibility pays off out of proportion to its size.

ChatGPT with browsing reaches a huge installed base, but intent varies more. It pulls from Bing, so Bing SEO matters here specifically [9]. Plenty of businesses ignore Bing, which means competition for Bing-sourced ChatGPT citations is thinner than for Google.

Claude has more limited web retrieval in its base product. Optimising for Claude is mostly about training-data presence, which you cannot directly control.

Gemini draws from Google's index, so your Google SEO gains carry over.

On a tight budget, spend on Google AI Overviews first. Then Perplexity, if your audience fits. Bing optimisation for ChatGPT citations is the underused opportunity almost nobody is working.

How do you vet an AEO or GEO service provider?

This matters more than usual right now. The barrier to calling yourself an AEO expert is roughly zero. Anyone can rebrand their SEO deck with new terms overnight.

Ask these before you sign. Can they show you a concrete measurement method, meaning which prompts they run, in which engines, at what frequency, and how they handle the noise (the same prompt returns different results on different runs)? Vague answer, vague reporting.

Ask for query clusters where they lifted citation rate, with the before and after numbers and the timeline. Good providers have this. They may not name clients, but they should have anonymised data ready.

Ask them to separate their Google AI Overviews work from their ChatGPT and Perplexity work. These need partly different tactics. If a provider treats every engine the same, they are not thinking hard about the problem.

Ask what happens when an engine updates its retrieval or ranking logic. The field moves fast. No update protocol means stale recommendations by month three.

Ask about their measurement tools. Purpose-built AI visibility SaaS, manual prompt testing, or something custom? Manual testing at low scale is fine for a one-off audit. It falls apart for monthly monitoring of a large brand. Tools in this space, including platforms covered on ai mode seo tool, are mature enough that any credible provider should be running one.

Spawned offers an AI visibility audit that documents your current citation rate across the major engines before any work starts. That baseline is what lets you hold a provider accountable later.

What content changes actually improve AI citations?

From the GEO research and the practitioner evidence so far, a handful of changes carry most of the weight.

Factual density. AI systems cite sources that back the specific claim in the user's query. A page that makes one clear claim per section, each with a real number or named study attached, gets cited more than a page of the same length spread across vague assertions. Aim for at least one extractable fact every 150 to 200 words: a statistic, a date, a named study, a defined threshold.

Author credentials and E-E-A-T signals. Google's Search Quality Rater Guidelines define E-E-A-T as Experience, Expertise, Authoritativeness, and Trustworthiness [2]. Built for human raters, these signals also seem to feed AI retrieval. Pages where the author has a verifiable professional background, links to their credentials, and gets cited elsewhere on the web perform better.

Structured data markup. FAQ schema, HowTo schema, and Speakable schema help AI parsers find the question-answer pairs and read-aloud-friendly segments in a page. Google's Speakable documentation says the markup helps Google Assistant and voice-enabled surfaces identify which parts of a page suit reading aloud [5]. The same logic carries to text-based AI answers.

Citations inside your content. Linking out to primary sources (government data, peer-reviewed research, official agency guidance) raises your page's perceived authority. The GEO study found adding authoritative quotes lifted citation by up to 30 percent [1]. Citing strong sources and being cited by them both help.

Content format. Direct question-and-answer structure matches how engines retrieve. A page that states a question, answers it immediately, then adds supporting detail gets cited more reliably than one where the answer hides in paragraph four.

What are the risks and honest limitations of AEO/GEO services?

The main risk is paying for work you cannot measure. Without a baseline citation rate before the engagement and consistent tracking over time, you have no way to know whether the money is doing anything.

Model opacity is structural. AI companies do not publish their retrieval or ranking criteria the way Google has published ranking guidance over the years. OpenAI has no public equivalent of Google's Search Quality Rater Guidelines. Perplexity's ranking signals are undocumented. So AEO/GEO recommendations are partly inference from observed behaviour, not confirmed from first-party sources. Anyone who tells you otherwise is guessing with confidence.

Model updates can wipe out gains. A provider can double your Perplexity citation rate in a quarter, then an update changes retrieval logic and it drops. Not necessarily their fault. It does mean you should budget for ongoing work, not a one-time fix.

Attribution is genuinely hard. A user asks ChatGPT about project management software, ChatGPT recommends your product, the user then Googles your brand and converts. That conversion lands in your analytics as direct or branded search, never as an AI referral. Most attribution models undercount the revenue from AEO/GEO work by design, and nobody in the industry has solved it yet.

And some of what gets sold as AEO is just good content marketing wearing a new badge. That is fine, because good content marketing is valuable. But do not pay a 40 percent premium on content costs for an AEO label if the underlying work is identical.

How should a company build an internal AEO/GEO capability versus hiring an agency?

For most companies under $50 million in revenue, a fully internal AEO/GEO team does not make sense yet. The practice shifts too fast, the tooling is immature, and the talent pool is thin. The better setup is a lean internal point person (usually a senior content or SEO manager) who owns strategy and holds an agency or tool vendor accountable for execution.

For larger enterprises, the math changes. If you publish 50 or more content pieces a month and your brand surfaces in AI queries thousands of times a day, the monitoring and optimisation volume justifies in-house tooling and a small specialist team.

Three internal capabilities are worth building regardless of size. A prompt-testing protocol, meaning a fixed set of queries you run across engines every month. A basic AI visibility measurement stack, whether a SaaS tool or a disciplined spreadsheet process. And editorial standards for factual density and structured formatting that apply to every new piece of content.

Spawned's platform handles the measurement layer for teams that would rather not build monitoring infrastructure themselves. A demo shows you your current baseline before you spend a dollar on optimisation.

For where the AI search market is heading, the ai search news feed is worth bookmarking.

What does a good AEO/GEO strategy look like in practice?

A realistic six-month programme for a mid-market B2B company runs something like this.

Month one is the visibility audit. Run 50 to 100 representative queries across ChatGPT, Perplexity, Gemini, and Google AI Overviews. Document where you appear, where competitors appear, which sources get cited most, and where the content gaps are. Two to three weeks of work, and it produces a prioritised list of opportunities.

Months two and three are content and schema. Reformat high-traffic pages to front-load direct answers. Add FAQ schema. Add author credentials. Raise factual density with real statistics and citations. Produce four to six new pieces aimed at the highest-priority question clusters the audit found.

Months four and five are authority building. Earn citations on third-party sites the engines already treat as authoritative in your category. This is the digital PR side of AEO. It is slow, but the GEO research suggests it moves citation frequency in a real way [1].

Month six is measure and iterate. Re-run the exact query set from month one. Compare citation rates. Find which changes moved the needle. Double down on those, cut what flatlined.

That is the shape of it. The specifics change with your industry, with which engines matter to your audience, and with how much your existing content overlaps the question clusters your customers actually use. There is no universal playbook yet. That is exactly why good providers earn their fee and why cheap, templated AEO services keep underdelivering.

Sources

1. Aggarwal et al., 'GEO: Generative Engine Optimization', arXiv (Princeton/Georgia Tech, 2023)
2. Google, Search Quality Rater Guidelines (2023 edition)
3. OpenAI, official blog
4. BrightEdge, 'AI Search: The State of SEO Report 2024'
5. Google Developers, Speakable schema documentation
6. Seer Interactive, AI Overviews frequency study 2024
7. Perplexity AI, official site
8. Google, 'How Google Search Works' documentation
9. Microsoft, Bing Webmaster Tools
10. Google Developers, structured data documentation