Method
How Pico Neural Measures AI Visibility
Pico Neural measures AI visibility engine by engine, as rows of observation with a stated limit. Five engines are named in the public pages; “outside” means not named, not excluded.
To see how Pico Neural measures AI visibility, start with a concrete check: take a question a real customer would type, then compare the answer you get with the one Pico Neural records. The public pages name five engines: ChatGPT, Perplexity, Gemini, Grok and Google AI. Claude and Grok are measured only after you connect them. When we say an engine is “outside” what Pico Neural observes, we mean it is not named in the public pages, not that it is excluded. The register is clear on this point: what is not named is not described. That distinction is the first step toward reading observations without turning them into guarantees of position.
Five named engines and a boundary for reading
AI visibility in Pico Neural is not a single score and not a promise of ranking. It is a set of observations about answers, read engine by engine. ChatGPT, Perplexity, Gemini, Grok and Google AI are the engines named in the public pages. Claude and Grok are measured only after you connect them: no connection, no observation. Gemini and Google AI Overviews are not the same observation, even if the names sound close. Read them as two separate rows, each with its own result.
An API response is not a person's screen. That is a limit on reading, not a technical footnote: what we observe through an API can differ from what a user sees in their own context. Read observations as indications of how answers behave, not as a snapshot of any individual experience.
How to read a row of observation
What follows is a Pico Neural working method, not a universal standard: fill in one row for each engine-and-question pair, with the result and the limit stated plainly. Here is a completed example to show the format:
- Engine: Perplexity
- Question: “how is AI visibility measured”
- Result: the product name appears in the answer, with no direct link
- Limit: API observation, not a user screen; a single reading, not a time series
The completed row shows why position is the wrong word: the result describes what the answer contains, and the limit says what that row cannot prove. Drop the limit and the result reads as a guarantee. Keep it and the result stays an observation.
Four steps to measure without misreading
- Pick the question and write it the way a customer would. Check: the question does not contain your product name and is not an internal search phrase. If it contains the product name, you are observing an answer that is already steered, not a discovery.
- Name the engine from the named list. Check: the engine is one of ChatGPT, Perplexity, Gemini, Grok and Google AI. If the engine is not named in the public pages, mark the row “not named,” not “excluded.” Claude and Grok require a connection: without one, the row does not exist.
- Record the result descriptively. Check: the result says whether the name appears, in what form and in what context, without words like “first,” “best” or “guaranteed.” If you cannot describe the result without position adjectives, rewrite the row.
- Write the limit next to the result. Check: the limit states whether this is an API response or a user screen, and whether Gemini and Google AI Overviews were read as separate observations. A row without a limit is a claim, not an observation.
The point of this piece is a distinction that often gets lost: AI visibility in Pico Neural is not a list of covered and uncovered engines. It is a set of observation rows, each with a declared limit. The second half of the reader's question, which engines stay outside, has to be read in the register's own terms: “outside” means not named in the public pages, not excluded. That difference changes how you use the observations. Read “outside” as “excluded” and you tend to treat the observation as complete. Read it as “not named” and you know the described perimeter is the five named engines, and that Claude and Grok join only after a connection. In practice, the row filled in with engine, question, result and limit is the format that keeps an observation readable without turning it into a guarantee. The four-step method is a Pico Neural working proposal, not a universal standard: it keeps apart what was observed, what was read through an API, and what is not named. That separation is the real content of measurement: not a number, but a row that says what we know and what that row cannot prove.
A limit you can check
The limit is verifiable in the row itself. If the engine is missing from the five named ones, or the connection note for Claude and Grok is missing, or the difference between an API response and a user screen is missing, the row cannot be read as an observation. Likewise, a row that treats Gemini and Google AI Overviews as the same observation has to be split in two. “Not named” does not mean “excluded”: that is the boundary that keeps the reading honest.
For the full picture of how measurement works and how to read observations, see the in-depth page.