Search is no longer a list of links. In August 2026, 80% of the searches where Flowup ranks in Brazil already showed an AI Overview (Semrush, Aug 2026). Ranking is still necessary; it is no longer sufficient.
By Guto Bertoncini, founder and technical lead of FlowupUpdated on 18 min read
5.0 on GoogleSince 2011Proprietary B.I.N.A. Method
Summary
The overlap between the sources cited in AI Overviews and the organic top 10 fell from 76% to 38% in seven months (Ahrefs, 2026). The two disciplines are compared in the GEO vs SEO guide.
What you get
At Flowup, GEO is not a package of posts: it is an authority engineering project with six named deliverables, defined timelines and continuous measurement. The entry point is the B.I.N.A. Diagnosis, which audits how your brand appears in AI answers today and sizes the project before any larger contract.
A prompt panel across ChatGPT, Gemini, Perplexity and AI Overviews: where your brand appears, how it is described and who is taking its place today.
An architecture of pillar pages, clusters and a glossary that organizes the company's knowledge for people, search engines and AIs.
The brand's official knowledge base on its own domain: canonical facts, numbers and positioning in an extractable format. It is the document you want the models to use when they talk about you.
The technical layer for machine reading: graph-based JSON-LD, connected entities, llms.txt and AI crawlers allowed by explicit rule in robots.txt and in the firewall.
Definitions, sourced statistics, tables and FAQs signed by identifiable experts: machine-readable E-E-A-T, in the pattern the Princeton study measured as more visible in generative engine responses.
Share of answers: a recurring prompt panel, a log of citations, AI referral in analytics and correlation with pipeline. A monthly report, with no promise of guaranteed citation.
Three published Official AI Knowledge Bases: open them, read them and compare them with what you have today.
AI knowledge basesSCA Brasil Official AI Knowledge BaseOfficial repository for AI: primary sources, usage guidelines and a citation model.
AI knowledge basesBioenergia Brasil Official AI Knowledge BaseSource governance for AI: what to cite, from where, and which entity this is.
AI knowledge basesRewood Official AI Knowledge BaseCanonical pages for AI on the Rewood website and on the book hub.
Method
The B.I.N.A. Method organizes the project into four layers. Each one has its deliverables and a completion criterion, and none skips the one before it, because content published on top of a confused entity does not build authority, only volume.
Diagnosis of the digital entity: how the company is described on its website, in external profiles, in search results and in AI answers. Next comes the Semantic Information Base: pillar pages, clusters, a glossary and the Official AI Knowledge Base on the company's own domain.
See layer B in the methodMapping of the questions, comparisons, objections and sub-queries that an AI derives from a prompt. With that, we build the prompt panel across ChatGPT, Gemini, Perplexity and AI Overviews, monitored from the first month to the last.
See layer I in the methodCitable content with authorship: definitions, sourced statistics, tables, FAQs and direct answers signed by experts, connected by internal links and backed by evidence, case studies, external references and earned mentions.
See layer N in the methodThe technical layer for machine reading: graph-based JSON-LD, connected entities, llms.txt and AI crawlers allowed by explicit rule. On top of that comes the monthly measurement of AI presence: share of answers, logged citations and AI referral in analytics.
See layer A in the methodResults
Three public projects in Brazil, with metrics from stated sources. In all of them, the canonical layer is the Official AI Knowledge Base; the numbers come from Search Console and Semrush, with the period and the cutoff criteria published in each case study.
Construction, engineered wood
Fuels, ethanol
Trade association, bioenergy
In every case study, the SERPs show AI Overviews; none claims guaranteed citation inside each answer. It is the same yardstick we apply to ourselves, right below.
Our own yardstick
A GEO agency has to accept its own yardstick. These are Flowup's numbers in search and in AI, measured in Brazil, with source and date. The AI Overview presence counter is updated every month on this page.
We apply to ourselves what we sell: the same prompt panel, the same share of answers and the same monthly report we deliver to clients.
Investment and timeline
The investment is sized by three variables: the number of topic territories the brand needs to own, the technical state of the website and the volume of content to produce or rewrite. That is why the starting point is always the B.I.N.A. Diagnosis.
It audits AI presence, entity and citability. You receive a prioritized plan, with scope and price, before any larger contract.
Timelines follow the pattern of the projects published in the case studies. Projects with a sound technical base and existing content to restructure move faster than projects that start from scratch.
GEO is asset building, not a campaign. What you pay for is the engineering of a source the AIs come to trust. What you measure is its presence, month by month.
Next step
Is your company ready to be understood by the AIs?
The next step is the B.I.N.A. Diagnosis: how your brand appears, how it is described and how it can be strengthened in search engines, answer engines and AI systems. By WhatsApp or by email, with a prioritized plan and a defined scope.
FreePrioritized planNo commitment
Take the free B.I.N.A. DiagnosisDefinition
GEO, or Generative Engine Optimization, is the strategy of optimizing a company's information presence so that generative engines and AI systems understand, synthesize, recommend and cite its information.
The core logic of GEO
The clearer, more consistent, more specific and more verifiable a company's information base is, the easier it is for search engines and AIs to treat it as a source.
In practice, GEO combines technical SEO, semantic architecture, structured content, source authority, organized data and direct answers to increase the chance of a brand appearing as a reference in AI-generated answers.
This does not eliminate SEO: while traditional SEO makes sure the page is found, GEO works so that it is used in the answer. Crawlability, indexing, performance, internal links, headings, structured data and useful content remain the foundation. What changes is that optimization now also considers generative synthesis, direct answers, citations and entities. The full definition and the checklist are in the complete guide to Generative Engine Optimization.
The canonical layer disambiguated the acronym SCA: 43 keywords in the Top 10, up from 8 a year earlier.
See how it was doneTerms in this sectionGEOAEOAI SEOEntityAI OverviewsStructured datallms.txtOfficial AI Knowledge Base
How it works
Each platform has its own models and criteria, but the pattern is shared: understand the question, retrieve useful information, synthesize an answer and, when the product allows it, present the sources. On Google, the Google AI features (AI Overviews and AI Mode) may use query fan-out; ChatGPT Search and Perplexity operate as AI search, a conversational search supported by sources.
The question is broken down into related sub-queries and the system looks for candidate documents in indexes and search engines. Without a crawlable base, the game ends here.
The engine chooses specific passages, not whole pages, prioritizing clarity, specificity, verifiable data and consistency across sources.
The answer is generated anchored in the selected passages, combining entities, relationships and evidence from the retrieved sources.
The system decides which sources to display as citations. Each platform attributes differently; diversifying signals is safer than betting on a single factor.
Implication for companies: vague, disorganized or isolated pages do not provide enough context to get past stage 2. A semantic information base helps the AI understand the relationship between topics, services, evidence and authority.
Comparison
GEO does not replace SEO and does not compete with AEO. The three disciplines complement each other in building an organic presence prepared for the new search.
| Discipline | Main focus | How it contributes | Practical example |
|---|---|---|---|
| SEO | Visibility in search engines, crawling, indexing, intent and organic rankings. | Provides the technical and editorial base for pages to be found, understood and evaluated. | An indexable, fast pillar page with internal links and useful content. |
| AEO | Optimization for direct answers in answer engines, snippets, assistants and FAQs. | Turns questions into clear, short, citable answers that are easy to extract. | Objective answer blocks for questions such as “what is GEO?” or “does GEO replace SEO?”. |
| GEO | Presence in AI-generated answers, syntheses, recommendations and citations from generative engines. | Organizes the brand's semantic authority so that the AI understands the full context. | Clusters, pillar pages, evidence, definitions, structured data and consistency across entities. |
The New SEO is the integration of these layers. It stops being a technical checklist and becomes a strategy of semantic authority, informational clarity and long-term organic assets, described in AI SEO.
Who it is for
The application changes with the segment, but the logic is the same: turning strategic knowledge into a clear, organized and trustworthy source.
Needs to explain complex solutions, comparisons, integrations and selection criteria to appear in consultative research and generative answers.
Example: HPROJ case study
Needs authority built on educational content, authorship, trust criteria and responsible answers to sensitive questions.
Example: Dr. Ana Vega case study
Needs to be the official source the AIs return for the name of the sector, with citation governance and disambiguation of similar names.
Example: Bioenergia Brasil case study
Needs to structure categories, buying guides, attributes, comparisons and reviews to influence buying journeys.
Example: Mantiqueira case study
Needs to spell out specialty, method, experience, area of practice and the answers to the audience's recurring questions.
Example: Dr. Alexander Kopelman case study
Needs to fix vague descriptions, disconnected pages and inconsistent messages before increasing editorial volume.
Strategy
A good GEO strategy does not start with content volume. It starts with architecture, intent, context and authority, and only then defines the optimization strategies for each page.
Who the company is, what it does, for whom and with what evidence, described identically on the website, in external profiles and in structured data.
Informational, comparative, commercial and transactional questions that an AI can break down into sub-queries.
Objective passages, definitions, lists, tables and direct answers that make the information easier to extract and to reuse responsibly.
Sources, authorship, references, track record, case studies, research and editorial consistency that sustain trust.
Pillar pages, clusters, glossary and commercial content connected in an understandable semantic network.
Structured data, headings, breadcrumbs, schema and accessible text that help engines interpret the page.
Clearer pages, organized direct answers, less ambiguity and technical preparation for crawling, indexing and understanding.
Stronger pillar pages and clusters, internal links, topical authority and presence in informational and comparative searches.
An organic digital asset, entity recognition, consistent presence in AI-assisted journeys and less dependence on paid media.
Beyond the results, I would highlight the clarity of the strategy, the organization of the work and the long-term vision Guto brought to the project.
Companies that do not organize their information base let third parties, competitors and generative models interpret the brand from incomplete signals. The most frequent risks:
Measurement
GEO should not be sold as a guarantee of AI citation. Generative platforms are dynamic and personalized, test different models and vary their answers according to context, location, history, language and index updates.
Even so, the strategy is tracked through practical indicators:
The full protocol is in how to measure AI visibility.
Qualified organic traffic measured in Search Console: 19,000 organic clicks in 359 days.
See the case study numbersThe most important metric is not showing up in an AI answer once. It is increasing the company's capacity to be understood, trusted and chosen in journeys increasingly mediated by answer engines.
We have been Flowup's partners for more than 10 years and, over that time, our trust has only grown stronger. The service is excellent and the commitment to our requests is total.
Flowup quickly understood my company's positioning and complexity, translating concepts into clear, strategic, high-level presentations.
The website now delivers something far more strategic, organized and aligned with what we wanted to convey.
Frequently asked questions
The questions that come up most often in meetings and in searches on the topic.
Is your question not here? The B.I.N.A. Diagnosis answers with your brand's data.
Take the free B.I.N.A. DiagnosisTransparency
This page uses primary sources to support the technical concepts about search, AI, structured data and answer engines. Flowup's own statistics carry their source and date in the text itself.
Flowup Agency. GEO agency (Generative Engine Optimization): authority engineering so your brand can be cited by AI. Guto Bertoncini, updated on . flowup.agency/en/geo/
Does your brand already appear in AI answers?
Request the B.I.N.A. Diagnosis and find out how your company is described today by ChatGPT, Gemini, Perplexity and AI Overviews, and what is missing for it to be cited.
5.0 on Google · Since 2011 · Case studies with stated sources
