How your company becomes visible in ChatGPT, Perplexity and Google AI Overviews

Three abbreviations, one goal: being cited as a source inside the answers of AI systems. The terms sound similar but mean different things. If you discuss the topic with agencies or internal teams, it helps to keep them apart.

AEO, Answer Engine Optimization, describes the practice of preparing content so that AI-driven answer engines such as ChatGPT, Perplexity and Google AI Overviews draw on it as a source. AEO is the operational term marketing leads use day to day. The focus is one question: how do I get my company cited when a potential customer asks an AI something?

GEO, Generative Engine Optimization, is the academic term, coined in 2023 by a research team from Princeton University and IIT Delhi. The study was published in 2024 at the ACM KDD conference in Barcelona (DOI 10.1145/3637528.3671900) and is the first peer reviewed research on the subject. GEO describes the systematic optimisation of content for generative search engines, including measurable methods such as Quotation Addition at plus 41 percent, Statistics Addition at plus 31 percent and Cite Sources at plus 27 percent, rising to plus 115 percent for pages ranked fifth among the retrieved sources.

LLMO, Large Language Model Optimization, is a synonym for GEO used in some agency circles but absent from the academic literature. When someone says LLMO, they usually mean GEO.

In short: AEO is the practice. GEO is the science behind it. This guide uses both terms because most search queries mix them.

You are not alone in this. The German digital association Bitkom published two studies in 2026 that together sharpen the picture. The representative survey Digitalisation of the Economy 2026 (604 companies with 20 or more employees interviewed by telephone, calendar weeks 2 to 6 of 2026, published on 11 March 2026) shows that 41 percent now use AI against 17 percent the year before, with a further 48 percent planning or discussing adoption. At the same time 51 percent struggle to cope with digitalisation. The second study, Marketing in Digital Transition 2026 (180 companies from the digital sector within the Bitkom network, calendar weeks 44 to 50 of 2025, published on 12 February 2026 and marked by Bitkom as not representative), has 84 percent naming AI as the most influential trend through the end of 2027. As internal obstacles, 35 percent name a missing AI strategy and 34 percent poor integration.

The gap is real, but it is not permanent. Companies that build an AEO strategy now occupy positions in AI answers that will be hard to win later. Brands cited today settle in as the default answer. Brands that wait will be competing against an order that has already formed.

Until 2024 there was no scientific basis for the question of how to optimise content for generative search engines. The recommendations came from agency blogs, SEO social media and the anecdotal reports of individual vendors. That changed in August 2024 at the ACM KDD conference in Barcelona.

A research team of six scientists (Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande) from Princeton University and IIT Delhi published the study "GEO: Generative Engine Optimization" (DOI 10.1145/3637528.3671900, pages 5 to 16 of the conference proceedings). They tested 9 different optimisation methods across 10,000 search queries on several generative search engines.

The results overturn several assumptions the SEO industry has treated as settled for years. The strongest single method is Quotation Addition, the inclusion of quotes from credible sources, at plus 41 percent. Statistics Addition, meaning the inclusion of concrete, verifiable figures, delivers plus 31 percent, and Cite Sources on its own plus 27 percent. The metric is Position-Adjusted Word Count, the share of words in the AI answer weighted by position.

Cite Sources is a special case. When all competing sources are optimised at the same time, the method lifts pages at rank 5 of the retrieved sources by 115 percent, while reducing the visibility of the top-ranked source by 30 percent. So you do not need a top 3 ranking to appear in AI answers. And whoever already ranks first does not automatically win here.

The most surprising finding: keyword stuffing hurts. The method that has underpinned many SEO strategies for two decades produces a decline of 8 percent against the baseline in generative search engines. In the check against the live system Perplexity the drop was slightly larger at 10 percent. That is not a marginal effect. It is a change of paradigm.

Every AI platform selects sources by its own criteria. What works in ChatGPT does not automatically work the same way in Perplexity. Here are the platform-specific patterns, based on the most recent studies available.

ChatGPT delivers 87.4 percent of AI referral traffic (Conductor, 13 November 2025, analysis of 3.3 billion sessions and 13,770 domains from its own enterprise client base). ChatGPT favours content with concrete statistics, structured data and clear source attribution. In the Ahrefs correlation analysis across 75,000 brands (August 2025), brand web mentions correlate at 0.664 with visibility in AI Overviews, backlinks at only 0.218. Ahrefs stresses that no causation follows from this and that all values are moderate to very weak relationships on the Spearman scale. The direction is nonetheless clear: being mentioned counts for more than being linked.

Perplexity links every answer directly to its sources. Perplexity favours current content with clear authorship and verifiable facts. In German-speaking Europe, Perplexity is still smaller than ChatGPT but growing quickly.

Google AI Overviews appear in 25.11 percent of all searches (Conductor, The 2026 AEO/GEO Benchmarks Report, 21.9 million analysed queries). In healthcare the figure is 48.7 percent. Context from the same source: AI referral traffic so far accounts for an average of 1.08 percent of total website traffic and grows by around one percent per month. Visibility is therefore shifting considerably faster than measurable traffic. Building citations today means positioning for a channel that is still growing.

AirOps published a complementary study in April 2026: comparison pages with three tables receive 25.7 percent more AI citations, validation pages with eight list sections up to 26.9 percent more, and shortlist pages averaging ten words or fewer per sentence 18.8 percent more. The figures apply per page type and exclusively to ChatGPT; they do not transfer to other platforms.

Here is what makes it difficult: the loss caused by missing AI visibility never appears on an invoice. No debit, no red entry in the accounting system. Only the creeping sense that enquiries are thinning out and that competitors somehow seem to be everywhere. Behavioural economics calls this loss aversion, documented by Daniel Kahneman and Amos Tversky in Prospect Theory in 1979 (Econometrica, Vol 47 No 2, DOI 10.2307/1914185), for which Kahneman received the Sveriges Riksbank Prize in Economic Sciences in 2002. Losses only feel like losses when they arrive as a bill. Silent losses hurt less in psychological terms and are therefore tolerated for longer, even though they cost exactly the same.

The Seer Interactive data shows the effect concretely: brands cited in Google AI Overviews reach an organic click-through rate of 0.70 percent against 0.52 percent for brands not cited on the same queries, roughly 35 percent more. For paid ads the difference is 7.89 against 4.14 percent. Seer itself qualifies this by noting that no causation can be derived: it is equally possible that brands with stronger authority and higher baselines are simply cited more often. Read the other way round it still holds: whoever is not cited carries the full decline.

At ALVÉRA, Performance Solutions and Creative Studio cover the two sides of the AEO equation: the strategy behind it and the technical implementation in front of it. AEO is not a content project. It is an architectural decision that addresses schema, structure, source work and entity building at the same time.

Our scope covers five building blocks, put in place in this order. First, an AI citation audit that analyses where your company stands today in ChatGPT, Perplexity, Gemini and Google AI Overviews. Second, a GEO content architecture with pillar pages, FAQ structures and answer-first formats following the Princeton methodology. Third, schema markup implementation covering Article, Organization, FAQ, Speakable and Service schema with concrete offers. Fourth, AI crawler configuration in robots.txt for GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and GoogleOther. Fifth, monthly citation monitoring with concrete KPIs: citation rate, citation recall and AI referral traffic.

What we do not do: produce generic SEO copy and hope AI cites it. That does not work. Every page needs verifiable data, correct schema markup and real signals of experience and expertise. The reason our own pillar pages appear in AI answers is exactly this architecture. It is the same approach we implement for clients.

Not for companies that only want quick rankings. Paid search exists for that. AEO builds positions that hold, but there is no guarantee of visibility. In a Semrush measurement across 81 newly published pages, a large share was still cited by no platform after 30 days.

Not for very small companies under 10 employees with a tightly limited content budget. GEO needs real content with real data. If you have no content worth citing, even the best schema architecture will not help. In that case, invest first in your website foundation and then in AEO.

Not for companies unwilling to fill their content with real data, sources and expertise. AI recognises generic content and rates it as less worth citing. The Princeton study shows it plainly: keyword stuffing hurts, statistics and sources help.

What we do well: mid-sized companies in Europe that have or want to build a solid online presence and are now taking the step into AI visibility. In the Bitkom survey of February 2026, 84 percent of the digital companies surveyed named AI as the most influential trend through the end of 2027.