How we measure
Standing, the one number we promote.
Standing is ExcBrand's 0 to 100 measure of how the AI engines represent a leader · Sentiment, Visibility, Accuracy, Authority.
Standing is how strongly the engines describe you, for your name and your terms. It leads with sentiment, the tone of what is said about you, then adds whether you are visible, whether the facts are right, and whether you are cited as a source. We baseline it, then we raise it. This page says exactly how it is built, so a reader can hold us to it.
The four parts
Four parts, each from 0 to 100, with a fixed weight.
Sentiment leads. The other three follow, in fixed order, so the same read applies to every professional and every term. The weights sum to one.
- 01SentimentThe lead signal. How positively each engine frames you when your name comes up.×0.20
- 02VisibilityPresent in the answer. How often you appear when your name and your terms are asked about.×0.30
- 03AccuracyThe base it stands on. How correct the engines are about your record and your work.×0.25
- 04AuthorityWho vouches for you. How credible the sources the engines rely on for you are.×0.25
Standing = 0.20 × Sentiment + 0.30 × Visibility + 0.25 × Accuracy + 0.25 × Authority.
The engines sampled
Five engines, asked the same questions.
Each week, for the name and for each chosen term. The questions are the ones a board, an investor or a journalist would ask.
- 01ClaudeAnthropic. Asked about the name and each term; the answer and its sources are read.
- 02OpenAIChatGPT. Asked the same questions; the answer and its sources are read.
- 03GeminiGoogle. Asked the same questions; the answer and what it draws from Search are read.
- 04PerplexityAsked the same questions; its cited web answers are read source by source.
- 05GoogleThe results and the AI Overview for the name and each term are read.
Confidence
What the number does and does not claim.
- Measured or not measured
- A number that was not measured is shown as not measured. It is never a zero and never a sample. Every reading reports attempted and completed as two figures.
- A degraded read is labelled degraded
- When an engine does not answer inside its deadline, the reading for that engine is marked degraded and a partial audit never presents as complete.
- Weekly, then official
- A weekly reading shows direction. Real movement is measured monthly, and the monthly reading is the official number.
- No benchmark
- We publish no industry average, no percentile and no sample figure. A number on this site is a real read of a consenting record or it is absent.
How the loop reads it
Baseline, move, re-measure.
Each week we ask the five engines how they describe you, for your name and each chosen term, and we read the sentiment, not only the facts. We choose the strongest truthful angles, publish what you approve, and then measure how the engines change their answer.
You see what changed, in plain language, and which engine changed it. Readings belong to the professional: no client’s reading appears on this site without written consent, and no sample stands in for one.