← All services

Generative Engine Optimisation

Be the answer, not the tenth blue link.

Your buyer opens an assistant and asks who does this in Croatia. One name comes back with a citation. GEO is the engineering and editorial work that makes that name yours - measured, re-tested monthly, and reported in plain language.

Ask for a baseline

0

Number of clicks an AI answer sends you when it summarises your page without naming you.

4

Engines tested every cycle - the general assistants plus the search-native ones your market actually uses.

30d

Re-test cadence. Models change their minds; a one-off audit goes stale within a quarter.

The engagement

Step one - fixed fee

Visibility baseline

A prompt set built from how your buyers actually ask - in Croatian and English - run across the engines. You get the transcript, who gets cited instead of you, and a ranked list of what to fix. Two weeks.

Step two - monthly

Fix & re-test

Entity and schema work, answer-shaped content, citation placement where the models look, then the same prompt set re-run each month so movement is visible rather than argued about. Cancel any month.

What lands on your desk

  1. Prompt-set transcript

    Every question, every answer, every citation, dated. Your own evidence file, not a dashboard screenshot.

  2. Entity & schema map

    How machines currently understand your organisation, products and people - and the structured data that corrects it.

  3. Answer-shaped content plan

    The pages worth writing, in the order worth writing them, with the question each one is meant to win.

  4. Crawl & access report

    Which assistants can read you at all: robots rules, render-blocking, feeds, and the security layer that quietly blocks them.

  5. Monthly one-pager

    Movement, losses, next actions. Short enough that your board will actually read it.

Honest limits

What GEO is not

No one can buy placement inside a model's answer. Anyone promising a guaranteed rank in ChatGPT is selling you a screenshot.

Not a guarantee

You get measured movement and a documented method, not a promised position.

Not a replacement for SEO

Classic search still pays the bills. GEO sits on top of a technically sound site - if the foundation is weak, we fix that first.

Not content farming

Fewer, better pages that a model can quote with confidence. Volume is what got the web into this mess.

Questions I get

Is GEO just SEO with a new name?

No. Classic SEO optimises for a ranked list of links; GEO optimises for being selected, quoted and cited inside a generated answer. Different evidence, different fixes - though a technically healthy site is the precondition for both.

How do you prove it worked?

The same prompt set, re-run on a fixed cadence, with dated transcripts. You can see the exact answer that named you, or the competitor cited instead. No proprietary score.

Do you work in Croatian and English?

Both, and it matters here: models answer differently by language. Croatian prompts often surface a completely different set of businesses than the English equivalent, so both are baselined.

Can this run alongside my current agency?

Yes. GEO work is mostly structured data, entity clean-up and a content plan - it hands over cleanly, and I am happy to write the tickets for someone else to ship.

What if my site blocks AI crawlers?

Common, and usually accidental - a security rule or a bot filter. That is the first thing the crawl report checks, and it is often the single largest win available.

Find out what the models say about you.

[email protected]