How to read this
A single brand score is a flat reading of a thing that isn't flat. The Brand Spectrometer un-flattens it one step at a time. Move the dial below and watch the one number open into eight dimensions, then into the audiences that disagree, then into how they feel and how sure we are.
The detail dial · move it
Each step adds to the one before — nothing is swapped out. Every step also says, in plain words, what it still hides.
−1 · One number. A single brand-health score: every dimension and every audience averaged into one line. It's not wrong — it's just low-resolution. This is the world most metrics live in.
0 · Headline. That one number was a blend of eight dimensions. The line splits into eight. Already you can see the brand is read more strongly on some dimensions than others.
1 · Compare. Each of those was an average over audiences who disagree. The band is the spread; each line is one audience. The same brand, read differently by each.
2 · Sentiment. Height is how strongly a dimension registers; the dot on the line is the feeling — its valence, from −1 (negative, left) to +1 (positive, right). A line that's high but whose dot sits far left is the dangerous one: strong, and negative.
3 · Resolve. The halo around each line is the 95% confidence interval — how sure we are. And the instrument carries a noise floor: gaps smaller than the floor are reported as "can't tell," not as findings.
4 · Cloud. The line's weight is its emission strength — how loudly the brand emits on that dimension — and every signal source behind the reading is shown. The full distribution, not a point.
Why it's called a spectrometer
A real spectrometer splits incoming light into separate wavelength bands and measures each one. This does the same with a brand. Picture every public signal about a brand — a review, a post, a press piece, a survey answer — as a star in the sky.
Each column of the strip is the sky photographed through one filter. The "Semiotic" column lets through only Semiotic-wavelength light, so you see which stars shine on that dimension, and how brightly. Slide to the next column and you're looking through a different filter.
Here's the key: you can't find an audience from one filter alone. Through the Semiotic filter the stars clump one way; through Economic, another. An audience (a cohort) is the group of stars that stay together across all eight filters — a pattern in the full picture, not in any single band. That's why the same audience sits at a different height in every column: same group of stars, read differently at each wavelength. (That difference between audiences is the measurement — what the theory calls metameric variance.)
You choose how finely to group the stars. Loosely, you get a few broad audiences; finely, many small ones. The instrument doesn't impose the audiences — it shows the stars and lets you set the resolution.
A brand is completed in the observer
A constellation isn't a property of the stars — it's a pattern you see from where you stand. The stars (the signals) are real. The shape they make is the observer's. Move to a different place in the universe and the same stars fall into an entirely different figure. That different standpoint is a different audience. So there's no single "true" brand to find — only the pattern each audience completes, and (from where they stand) some stars are hidden from them altogether.
The halo around an audience's line is how tightly its attributable stars cluster: a tight halo means a confident, agreed reading; a wide one means the audience is unsure or divided.
The old way reads the sky like astrology: it flattens it to a 2-D dome and collapses a whole person to one sign — one number. Same sky; very low fidelity. The Brand Spectrometer reads it like astronomy with a spectrograph: it splits the light into bands and recovers what the flat map threw away — what each signal is made of (the eight dimensions), how it's moving (heritage and momentum, the Temporal band), how far apart audiences really are (the spread), and — honestly — what it cannot resolve (the noise floor). Climbing the dial is that whole move, from one flat sign to a multi-band reading. We're judging the instrument's fidelity, not anyone who reads stars.
What it deliberately does not show
No single "true" silhouette — there's no presumed ground truth, only a distribution across audiences. Anything below the noise floor is reported as unresolved, not zero: the instrument tells you when it can't tell two audiences apart, instead of inventing a difference. It measures audiences, never individuals.