When someone asks ChatGPT, Gemini or Perplexity for the best option in your category, the answer names three to five brands. We measure whether yours is one of them.

What an answer looks like

Solaria sells sunscreen in Mexican pharmacies. Its absence was decided by sources the brand does not control.

Marca inventada, competidores inventados, cifras inventadas. La estructura es real; los números no son de nadie.

The full report page, on the invented sample: the consumer sample one-pager.

The problem, in the buyer's words

Buyers stopped searching. They ask. The sentence "ChatGPT no recomienda mi empresa" usually arrives late, from a prospect who already chose a competitor.

The answer to a category question names three to five brands. The moment of loss is quiet: the competitor gets named, you do not. And the brands left out do not even find out they were left out. No report shows it. No alert fires.

You cannot fix what you do not measure.

What we do

We measure and improve your visibility in AI engines, with our own reproducible studies.

This is market research on AI answers ("esto es investigación de mercado sobre las respuestas de IA"), not another dashboard. Your team keeps priorities, not another subscription. Every claim we make about your visibility comes from a study, every study runs on our own tool, answer.show, and you can ask to see the raw output.

What gets counted

Four outcome rates, measured the same way in every response of every run:

1. Mentioned (mencionada). The brand appears in an answer to a question that never named it. The floor: presence in the category conversation.
2. Recommended (recomendada). The assistant actively suggests it. Being listed among eight is not being told to buy it.
3. Named as the best (señalada como la mejor). The answer distinguishes it instead of grouping it. The strongest and rarest form of presence.
4. With a purchase path (con ruta de compra). The answer says where to buy it. The closest honest proxy to purchase intent, with no analytics needed.

From the invented sample:

Marca inventada, competidores inventados, cifras inventadas. La estructura es real; los números no son de nadie.

Every rate prints its denominator. A percentage with no population behind it is a decoration.

Also measured in every study: competitor mentions, every cited source (deduplicated, ordered by who can actually move it), and refusals and evasive answers recorded as their own status. When an assistant refuses to recommend, that is information, not missing data.

How a study runs

Three steps.

1. We build your prompt bank. Real buyer questions in three layers: unbranded, brand-named, competitive. Themes come from documented demand: Search Console queries, Google autocomplete, People Also Ask, forums, reviews. Not from a consultant's whiteboard.
2. We measure across four AI engines, with repetitions. AI answers vary from run to run. Repetitions make the numbers stable enough to report.
3. We read every response and count the same things every time. Mentions, recommendations, cited sources. One codebook, every run, so the numbers compare.

The full method is public: /metodologia/.

Where the lever is

Same brand, same question bank, mention rate by assistant type (invented sample):

Kept in Spanish for the final site: "el awareness en la memoria del modelo es débil; la web abierta es la que está cargando la marca."

Marca inventada, competidores inventados, cifras inventadas. La estructura es real; los números no son de nadie.

Reading: what the model remembers from training is weak. The open web is what carries the brand. That is why the work after measurement is mostly editorial, retail content and PR, not a website migration.

How to read the split, and why repetitions matter: /metodologia/.

Proof

• The method is published. Bank construction, models, repetitions, what gets counted, and the limits: /metodologia/.
• Sample reports, both segments. A consumer one-pager (Solaria, sunscreen in Mexican pharmacies) and a B2B one-pager. Both carry the same disclaimer: Marca inventada, competidores inventados, cifras inventadas. La estructura es real; los números no son de nadie.
• We measure ourselves. Our own category, "agencia GEO México", measured and published like any client study. [VERIFICAR: pending first run]
• Own technology, open data. One of the few agencies in Mexico measuring AI visibility with its own tooling and raw output you can inspect.
• Google, quoted: "optimizing for generative AI search is still SEO" (Search Central, AI optimization guide, 2026: developers.google.com). The same guide names llms.txt, chunking and special schema as myths. We do not sell any of them.
• OpenAI, quoted: do not block OAI-SearchBot if you want pages considered for ChatGPT search; GPTBot governs training, not search (OpenAI Publishers and Developers FAQ: help.openai.com).

Request the study

What to unlearn

1. There is no position one in AI answers. Whoever sells you one invented it.
2. Most of the answer was written by someone else. Third-party sources, not your site.
3. The same question answers differently each time. One screenshot is an anecdote.
4. A topic can matter before anyone searches for it.
5. The first run is a baseline, not a fix.

Full versions and what each one replaces: /metodologia/.

Who it is for

Study size

No fixed package. Three knobs: questions x models x repetitions.

• Small end: a directional sample. Tens of prompts, fewer models, fewer repetitions.
• Deep end: around 100 prompts, four or more models, five repetitions.

Illustrative configuration, from the invented sample: 60 questions x 3 assistants x 3 repetitions = 540 responses read.

The size is chosen with you in the levantamiento (the kickoff research), and the price is agreed before the first prompt is sent. Published numbers: [PRECIO POR DEFINIR]. Structure and scope per tier: /precios/.

Start with the question, not the tool

The right first step is not software and not a package. It is the question your business needs to answer: does the category conversation include us, and who is in it instead?

Request the study. It starts with a conversation about that decision, not with a demo.

Request the study