The Haivoronskyi Framework: every GEO factor in a single table

TL;DR

GEO has grown dozens of factors: schema, llms.txt, mentions, reviews, speed, profiles, sameAs. In articles they come as one long list, and it is unclear what matters more and where to start. Viktor Haivoronskyi, founder of the agency F-47, organized them into a simple 2x2 matrix. Two axes, four quadrants, and the whole GEO picture fits on one screen.

How the Haivoronskyi framework is built

The framework is built on two questions. First: is the factor inside your site or outside it, which is the vertical axis, internal and external. Second: is it the technical layer, that is how machines read the site, or the content layer, that is what they actually read, which is the horizontal axis, technical and content. Their intersection gives four quadrants.

The core idea of the framework: visibility in AI appears only when all four quadrants work together. Most teams pump one, usually internal technical, they added schema and relaxed, while the other three sit empty. Below the quadrants are arranged by descending significance for promotion. Important: this is the order of impact, not the order of action, we will return to the sequence of implementation at the end.

External content: the main lever of AI visibility

This is the strongest and at the same time the most neglected quadrant. AI assembles its answer mostly from what is written about you outside your own site, so external content most often decides whether you get into the recommendation or not. It includes:

• PR, media, ratings, reviews, testimonials.
• YouTube, LinkedIn, Reddit and niche forums.
• Comparisons with competitors and expert quotes on third-party platforms.

The logic is simple: models trust what is confirmed independently. One spot in an authoritative best-in-niche list weighs more than ten self-praises on your own site. So this quadrant deserves investment first, but only after there is something to confirm, which is the next point.

Internal content: what AI can actually cite

This is the foundation you control completely. If external content is the votes for you, internal content is what they discuss. Without strong answer pages there is nothing for AI to cite, even if people write about you. It includes:

• Answer assets (pages for specific customer questions), FAQ, tables, comparisons.
• Case studies with numbers, pricing, guides.
• Internal links and original datasets that exist nowhere else.

The core principle here was put well by Dejan.ai: content that is convenient for humans is convenient for AI. Give a direct answer in the first lines, add FAQ, tables and comparisons so the model can easily pull a ready fragment. Your own data and case studies with figures work especially well, because they cannot be obtained from any other source.

Internal technical: without it you simply will not be read

This is not a growth lever but a checkpoint. If AI crawlers cannot enter the site or do not understand its structure, nothing else matters. The ceiling of this quadrant is lower than the content ones, but you have to pass it first. It includes:

• txt, llms.txt, sitemap.xml, indexation, site speed.
• schema JSON-LD, datePublished and dateModified.
• Author pages, About, contacts and policies as trust signals.

The minimum is this: open the site to the necessary bots in robots.txt, add sitemap.xml and correct Schema.org markup on key pages (Organization, Product, FAQPage, Article). Worth a separate mention is llms.txt, a new pointer file that tells AI which of your pages are the main ones. The standard is still young and does not affect Google rankings, but as a cheap step for AI it already makes sense to add.

External technical: so AI treats you as one company

This quadrant is about recognizing the brand as a single entity. If you are named differently across sources, AI may fail to link them into one company and not understand that it is about you. It includes:

• A single brand name and correct company profiles.
• sameAs between the site and official accounts, accessible directory pages.
• Consistent NAP data, correct links and descriptions.

The work here is not creative but about accuracy: the same name everywhere, filled-in and non-contradictory profiles, a sameAs link. It does not give explosive growth, but it removes the situation where part of your mentions simply do not count because AI did not understand it was you.

In what order to implement the framework

Significance and sequence are not the same thing. By impact, external content comes first, but you have to start the work with the technical layer, otherwise there will be nothing and no way to confirm. The working order is this:

1. Internal technical: open the site to bots, add schema, sitemap and dates. This is the checkpoint.
2. Internal content: build answer pages, FAQ, comparisons, case studies and pricing. Now there is something to cite.
3. External technical: a single brand name, profiles, sameAs, NAP. AI starts to recognize you as one company.
4. External content: PR, ratings, reviews, mentions, discussions. The biggest gain, but it works only on top of the first three.

Self-check list for the Haivoronskyi framework

Go through one point per quadrant. If the box does not get checked, that is your growth point.

• The site is open to AI bots in robots.txt, there is a sitemap.xml and llms.txt.
• Key pages have correct schema JSON-LD, publication and update dates are visible.
• There are answer pages, FAQ, comparisons, tables and case studies for real customer questions.
• The brand name is the same everywhere, directory profiles are filled in, sameAs is set up.
• Company data (NAP, descriptions) matches across all sources.
• People write about you outside your site: media, ratings, reviews, discussions.
• You appear in third-party lists and comparisons in your niche.
• You regularly check what AI answers about your brand and close the weak quadrants.

If not all quadrants are checked on the list, that is normal: almost no one has them all closed, and that is where the work begins. To see your own picture by the framework on real data and understand which quadrant is dragging you down, start with a free AI visibility audit from F-47.