GEO consulting (SEO for AI)
So ChatGPT, Gemini, Perplexity and AI Overviews name you when your customers ask about what you do.
Book a 30-min call →I measure AI visibility with my own tracker: every day it asks the same questions to ChatGPT, AI Mode and AI Overviews and stores which brands show up, which sites they cite and which searches they run. Before that I led SEO at Softonic and word.tips.
What I do
The AI decides who to name before it searches
When you ask ChatGPT for the best consultants in some field, it does not search and pick from what comes up. It first decides the names, and then runs searches against each candidate's site to check case studies, clients and track record. I have seen it in the fan-outs of my own tracker. If you are not on that initial list, it never gets to look at your site.
That is why GEO has two parts. One is being in the sources the model draws that list from: media, lists, profiles and people it already trusts. The other is making sure your site holds up to verification when it looks you up: case studies with data, who you are and what you have done, in text that can be read.
And on informational questions something else happens: the AI draws on official documentation and on whoever has original data. If ten pages say the same thing, it can cite any of them. If only you have the data, the citation is yours.
- 01 Many answers, not one I run the same questions many times per topic and platform. A single answer is noise; the share of answers where you appear is the data.
- 02 The fan-outs The searches the AI runs before answering. They show which brands it brings already decided and which sites it checks them against.
- 03 Which sources it cites in your sector Media, lists, directories and profiles that come up again and again. Those are the places where you need to be.
- 04 What it says about you Showing up is not enough: it can name you in order to rule you out. I separate the mentions that recommend from those that do not.
- 05 A consistent entity Your site, your profiles and what is said about you tell the same story: who you are, what you do and who for.
- 06 Whether your site holds up to verification Case studies, clients and track record in readable text, and AI bots that are allowed in.
Teaching
Over 10 years teaching master’s programs
- Master in Search Engines · Universitat Pompeu Fabra 2015 — 2024 · Link spam and authority metrics
- SEO Master · Webpositer Academy 2018 — present · Search Console
- SEO Master · BigSEO Academy Present · Web migrations and log analysis
- Technical SEO Master · Asdrubal SEO Present · Web migrations
- Online courses · Wontalia Python scraping, Google Tag Manager and web migrations, among others
- Dean Romero's SEO and monetisation course · Dispara tus visitas Dissecting Google's Quality Raters guidelines
Talks
Speaker at conferences and companies
- Everything you wanted to know about Google Sheets but were afraid to ask See slides
- In-house training: an introduction to log analysis See slides
- In-house training: JavaScript + SEO See slides
- SEO processes, methodologies and automations at Softonic See slides
- Programming for SEO: basic concepts and practical uses See slides
- An SEO framework for successful web migrations See slides
- Advanced scraping, or how to turn the internet into your database See slides
- SEO 101: an introduction to SEO See slides
How we work
From the first call to results
- Day 1
Call and context
What your customers ask before buying, who your competitors are and in which markets you care about appearing.
- Weeks 1–2
Baseline measurement
The questions for each topic, run every day in ChatGPT, AI Mode and AI Overviews until we have a reliable baseline.
- Weeks 3–4
Diagnosis and plan
Fan-outs, the sources the AI cites and what it says about you and your competitors. The site and mentions plan comes from there.
- Every quarter
Quarterly plan
Absolute key results, such as in how many answers it names you, and initiatives with their own metric.
- Every week
Weekly sprints
Site changes and mentions work, with the weekly board review.
- Every month
Measurement
We measure again against the baseline: where you appear, which sources cite you and what the AI says about you.
The first 3 steps happen once, at the start. From "Quarterly plan" onwards, the work becomes a steady flow that repeats.
It is the same methodology in every project: a business goal, a diagnosis by page type, absolute key results and a weekly review. I explain it in full in this post
Deliverables
What you take away
-
Baseline measurement
The questions your customers ask, run many times in ChatGPT, AI Mode and AI Overviews. A single answer tells you nothing; many do.
-
Fan-out analysis
Which searches the AI runs before answering, which brands it brings already decided and which sites it consults to verify them.
-
Source map
The sites, media, lists and profiles the AI cites in your sector. Those are the places where you need to appear.
-
A site ready to be verified
Entity, structured data, case studies and pages a model can read and cite without hesitating.
-
Mentions plan
Where to get people talking about you so you make the candidate list the AI holds before it searches.
How you can hire me
GEO diagnosis
Baseline measurement, fan-out and source analysis, and an action plan. A fixed project of about four weeks.
Monthly advisory
Continuous measurement, work on the site and on mentions, and a monthly report on what moves.
In ongoing work, you choose how involved I get
-
Guided consulting
Regular sessions with your team. I set the priorities, review what has been done and measure the results. Your team executes.
-
Hands-on support
On top of that, I review changes before they ship and I am in your Slack or Teams for day-to-day questions.
-
Inside your team
Almost like an implant: I join your tools, write and prioritize the tickets myself, take part in planning and do the SEO QA of what ships.
We adjust it to the project and it can change along the way: some projects start guided and end with me inside the team, and the other way round.
Fixed price for the diagnosis, based on the topics and markets we measure; the advisory is priced by month. I send you the proposal in writing after the first call.
FAQ
Frequently asked questions
Can't find your question? Email me at [email protected] and I'll reply within 24 hours.
Do I need an llms.txt?
Today there is no evidence it helps. In its guide to AI features, Google says no special file is needed to appear, and there is no record that it uses one. If you want one, it costs little, but it is not what will get you to show up.
How do you measure AI visibility?
By running the same questions many times and counting in how many answers you appear, because each answer is a different sample. I measure by topic, with fixed questions, in ChatGPT, AI Mode and AI Overviews, and I separate when it names you, when it links you and what it says about you.
How long does it take to see anything?
It depends where. What the AI pulls from searching the web (AI Overviews, AI Mode, ChatGPT with search) can move in weeks. What the model knows from memory only changes when it is retrained, and that takes months.
Do you guarantee that ChatGPT will recommend me?
No. Nobody controls what a model answers. What I do is measure where you stand, find which sources decide your sector, work to get you into them, and measure whether it works.
When shouldn't you hire me?
If you are looking for a trick to show up in ChatGPT next week. If your brand has no case studies, clients or presence anywhere yet. That comes before AI.
How much does it cost?
It depends on how many topics and markets we measure and how much site and mentions work we do together. I send you a fixed proposal after the first call.