Why your business doesn't show up in ChatGPT (and what a 0/100 actually means)

An accountant in São Paulo is looking for an agency to rebuild his website. He no longer types the question into Google - he asks an AI. That day, the answer mentions neither the agency down the road nor twenty-one other firms in his city.

What we measured

On 4 and 5 August 2026 we audited 24 Brazilian businesses - nineteen digital marketing agencies and five accounting firms - in Portuguese, using the questions a buyer actually asks before choosing. Each measurement covers two AI assistants (AEO, Answer Engine Optimization: the work of existing inside assistants' answers rather than in a list of links).

Twenty-two of them scored 0 out of 100. Only two stood out, at 32 and 17.

22 businesses out of 24 appear in none of the answers we measured.

And yet one of them had won a client through AI

The day after the measurement, one of those agencies wrote back, amused: it had just picked up a new client who arrived through AI. Its score, measured the day before: 0 out of 100.

Both statements are true at the same time. That is the most useful thing in this data, and the thing most often misread.

A zero is not a verdict of invisibility

A score of 0 says something precise: across the sample of questions we asked, on the assistants we queried, on that day, the business was cited in none of the answers. It does not say the business is absent from every possible question, or every engine, or tomorrow.

AI answers shift from one day to the next and from one phrasing to another. A business can be invisible on "best agency in São Paulo" and cited on a narrower question nobody thought to ask - which is exactly what happened to that agency. This is why an audit is a measurement, not a sentence, and why the trend over several weeks matters more than today's number.

The real signal is elsewhere: the ground is nearly empty

That 92% of a sample scores zero says less about those businesses than it says about the ground they stand on. In this country, in this trade, the space is still open. Almost nobody has started working on how they appear inside AI answers.

We promise no ranking - nobody can, and anyone who promises you one is selling something other than measurement. What we observe is that competition on this ground is currently thin, and that it will not stay that way.

What assistants actually cite

Assistants do not rank the way a search engine does. They cite what is structured, consistent, and corroborated across sources. In our measurements it is almost always the same families:

• the company's Google profile, complete and current
• reviews - and above all their consistency across platforms
• trade directories, including the ones everyone assumes are obsolete
• explicit service pages that name the trade and the city
• structured data, which tells the machine what it is reading

None of this is new, and most businesses already know how to do it. What was missing was never the method - it was the measurement: before, after, and every month.

Where to start

Start by knowing where you stand instead of guessing. The audit is free, with no account and no sales call: waseit.net/en/audit. You will see, question by question, what assistants answer about your trade in your city - and who gets named instead of you.

And if it comes back 0 out of 100, reread the section above: that is not a verdict. It is a starting point, with a date on it.