What each metric means

Every number on the dashboard is a share of runs, not a rank in a chart. A run is one prompt sent to one engine on one day, so a hundred runs is a hundred separate chances for your brand to be named.

Visibility Score

Runs where your brand appears in the top ten of the answer, divided by all runs.

This is the headline number and the one to watch weekly. It moves when your brand starts appearing in answers it used to miss, or when it climbs out of the tail of a long list into the part people actually read.

A brand new brand often sits in single digits. Anything above 30% in a competitive category means the engines genuinely know you.

Detection Rate

Runs where your brand is mentioned anywhere, divided by all runs.

Detection is always at least as high as Visibility. A large gap between the two, say 40% detection and 8% visibility, has a clear reading: the engines know you exist but keep putting you at the bottom of the list. That is a content and citation problem, not an awareness problem.

Top 3 Visibility

Runs where your brand lands in the first three positions.

The strictest of the three, and the closest to a business outcome: in a spoken answer or a short summary, position four onwards often never gets read out at all.

Average Position

The average position of your brand across runs where it was mentioned. Lower is better.

Only mentioned runs count, so a brand named once at position 1 shows a perfect 1.0. Read it next to Detection Rate, never on its own.

Share of Voice

Your mentions as a share of every brand mention in your answers.

This is the competitive number. If ten brands split a category evenly, each holds 10%. Yours tells you how much of the conversation you own, and the Competitors page shows who holds the rest.

Brand Sentiment

The share of your mentions that read as positive.

Measured on the sentence that names you, for your brand only. Competitor rows show no sentiment on purpose: the parser has nothing reliable to read there, and a fabricated 0% would look like a real measurement.

Citations

How often a domain is cited in the answers to your prompts.

Not a score for you, a map of the sources engines trust in your category. See Citations.

Why numbers differ from what you see yourself

Ask an engine the same question twice and you will get two different answers. Ours is the average of many runs across many days, yours is one run, and a single answer is not evidence either way. Also check that you are looking at the same market and the same engine: results differ sharply between them.

If the dashboard shows a floor of a few percent on a brand with no mentions at all, that is deliberate: a brand new brand reads its own zero as “this product found nothing”, when the honest reading is “there is work to do here”. The competitor table and per-prompt breakdowns always show raw values.