When an assistant does the shopping, your brand is the North Star.

What consumers are predicted to take away is not always what the work says

Across 612 brands the predicted consumer score and the brand score rank brands at 0.50. Consumer scores run about 16 points lower across the board, so the comparison is by percentile rank: a brand moves 23 percentile points on average. Predicted to rank higher with consumers than on their work (percentile, work → consumers): Wild Wing (0 → 86), CBC Gem (11 → 94), Meridian Credit Union (3 → 76), Church's Texas Chicken (14 → 85), Moosehead (7 → 77), Jack Astor's (14 → 83). Predicted to rank lower: Hoka (87 → 5), Nespresso (84 → 9), American Express (92 → 21), Nike (92 → 25), Ram (79 → 13), Bulleit (86 → 22).

If the prediction holds, the gap is the brief: where the consumer read ranks lower than the work, each piece is predicted to move people somewhere different even when it looks the part.

The consumer score is less reliable than the blind score (split-half 0.71 against 0.85) and has not been checked against real consumers at brand level, only ad by ad in banking. So the league stays ranked on what brands ship, and the consumer score sits beside it.

The market repeats your product and your website. It barely repeats your advertising.

Each row is how well one touch point, on its own, predicts where outsiders place the brand (1 = exactly, 0 = unrelated).

Advertising is the loudest thing a brand does and the thing the market repeats least, and AI assistants do not repeat it at all. Product is what gets talked about, and the website is what an assistant reads first. The score now weights touch points by how much the market repeats them.

Singular brands get described as themselves. Scattered brands get described as the category.

We asked three AI assistants what each brand is like and placed their answers blind on the Map, then scored how well each answer matches the brand’s own story, 0 to 100. For the most singular fifth of brands the match is 97. Matched against the same brands, the assistants’ descriptions of their category competitors score 79, so the assistant is describing that brand, not the aisle. For the least singular fifth the match falls to 64. 73% of top-fifth brands get a near-exact description, against 29% of the bottom fifth.

When the assistant speaks for you, a singular brand is repeated. A scattered one is paraphrased.

Singularity does not decide whether the assistant names you. It decides what it says when it does.

We told three assistants to just pick a brand in 87 categories, five times each: 1,305 picks. Whether a brand got picked did not follow its Singularity (within-category correlation 0.02; 914 of the picks named a brand in this league). We did not measure what the pick does follow. What changes with singularity is what the assistant says once it has named you.

Getting named is a different contest. Singularity decides what you are once you are named.