GEO audit and consulting
An AI assistant visibility audit: what they say about your brand today, why they say it and what to change so they say something else. You can start with a diagnosis, a full audit or ongoing support.
Let's talk about your project- 5
- levers reviewed in every project: access, understanding, entity, authority and reputation
- 10+
- years of average team experience in SEO, content and brand entity
- 48 h
- to reply, at most
Brands we've worked with
Do any of these situations sound familiar?
Almost nobody asks for a GEO audit out of curiosity. You ask when a decision needs judgement and the explanations on the table don't convince you.
You don't know whether you show up
Someone on the committee asks what ChatGPT says about the brand and nobody in the company has a data-backed answer.
The model describes you wrong
It credits you with products you don't sell, countries you don't operate in or prices that aren't yours.
It recommends your competitor
In the questions that decide a purchase in your category two other brands appear, and yours doesn't.
There's budget and no plan
The decision to invest in AI visibility is made, but nobody has ordered where to start or with what criteria.
You have a proposal and doubts
Someone has offered you AI visibility built on llm.txt and promises of appearing, and you want an independent second opinion before signing.
Traffic drops while answers rise
Impressions hold, clicks don't, and you want to know how much of that is the generative answer.
Four fronts we always review
The weight of each front changes with your category and how mature your SEO is, but none is left out: the reason you don't show up is usually in the one nobody looks at.
AI access and crawling
So the models' crawlers can reach, read and retrieve your content without stopping halfway.
See the detail
- Review of blocks from robots, WAF, CDN and bot rules in production
- Comparison between served HTML and final rendered content
- Log-based record of AI crawler visits when access exists
- Latency, errors and responses that return empty content
Extractable content
So an answer can cite you without having to interpret: clear data, self-contained statements and pages that answer.
See the detail
- Audit of the pages that should be sources in your category
- Structural rewriting: answer first, context after
- Concrete, verifiable data where there's marketing language today
- Coverage of the real questions asked before buying
Entity and consensus
So it's clear who you are, what you do and where you operate, and every source says the same thing.
See the detail
- Brand record: name, category, scope and proposition, unambiguous
- Structured data for organisation, product and authorship
- Consistency across site, profiles, directories and encyclopaedias
- Correcting what the model repeats wrongly about the brand
Third-party sources and reputation
So the sources the answer leans on include you and describe you the way you want to be described.
See the detail
- Map of the sources cited in your category today
- Presence in comparisons, listings, forums and reference media
- Reviews and opinions: what gets repeated and what ends up in the answer
- Priority on where it's worth being and where it isn't
Five questions you need to be able to answer
The whole audit is ordered around these five levers. If one fails, the rest don't matter.
- 01
Can they see you?
Technical accessibility for classic and AI crawlers: rendering, crawl budget, blocks and latency.
- 02
Do they understand you?
Semantics, structure and structured data. If a machine can't extract an answer from your page, it won't cite you.
- 03
Do they know who you are?
Brand centroid: which concepts models associate you with, and which ones they should.
- 04
Do they cite you?
Presence in the sources feeding the answers: media, forums, comparisons, directories and first-party data.
- 05
Do they trust you?
Reputation and consistency: what's said about you off your site weighs as much as what you say on it.
What an AI assistant really searches for
When someone asks about your brand, an AI assistant doesn't just search for your brand
It fires searches the user never sees, and uses them to decide which sources make it into the answer.
In-house study by MJ Cachón · GPT-5.5, August 2026 · 189 brand prompts
Read the full study- 69,8%
of searches go to open Google: your traditional SEO is still the foundation.
- 30,2%
use "site:" on a domain. If your site does not answer, a third party answers for you.
- 23,8%
look for exact phrases in quotes. Claims and figures become facts you must hold up.
Three projects, one model, one moment: these are patterns, not a market sample.
Who works on your project?
Senior profiles who have worked access, content and brand entity for years, not a team assembled last week to sell AI.
Why Laika?
Because we name the problem before selling the solution.
AI visibility is sold with a lot of promises. We deliver a dated record, a specific cause and a plan your team can execute.
Senior SEO, not a trend
What moves a generative answer is access, structure, entity and sources. We've worked on that for years, before it was called GEO.
Measured, not guessed
Every conclusion comes from a dated record of real answers, repeatable over time.
Honest about what isn't known
There's no official console and no guarantee of appearing. We say what can be asserted, what's a hypothesis and what's noise.
Genuinely prioritised
Ordered by what moves the business and by what your team can execute this quarter.
Frequently asked questions
Isn't this the same as an SEO audit?
It shares a foundation, but not a goal. An SEO audit chases rankings and clicks; this one chases being the source the answer cites and making sure what's said about you is right. The questions, the measurement and much of the work change — and a lot of it happens off your site.
Can this actually be measured?
Yes, with systematic question sampling and cited-source tracking, repeated over time. There's no official console, but there is method: that's why everything is delivered with a date and a screenshot.
Do you guarantee I'll appear in ChatGPT?
No, and be wary of anyone who does. What can be done is create the conditions for you to be considered, measure whether it shifts and correct. If your case has no room, we say so before starting.
Does it cover several assistants or just one?
The audit covers the assistants that matter to your business, usually ChatGPT, Gemini, AI Overviews and Perplexity. Afterwards you can go deeper on whichever carries most weight in your category.
How long until I see results?
The first findings and decisions land within weeks. Impact on visibility and revenue depends on your platform and your team's implementation speed: on mid-sized projects we usually talk about three to six months for solid movement.
I worked with another agency and it didn't work. Why now?
What usually fails is judgement, not effort: tasks get executed without deciding what matters. Before proposing anything we review your data, and if the problem isn't visibility we'll tell you, even if it means not selling the project.
Let's talk about your project
Tell us your category and your domain and we'll say, with no strings, whether you need a diagnosis, a full audit or ongoing support.
Want to know what assistants say about your brand?
We measure it with real questions from your category before proposing anything.
We look at your project before proposing anything. Reply within two working days at most.
Request a GEO diagnosis