The Brand Distinctiveness Index, Explained

The Brand Distinctiveness Index measures 1 property: whether a machine can tell your brand apart from your category. Not whether your branding is good, not whether the copy is clean, not whether the deck feels premium. Separability, scored 0 to 100, because separability is what AI engines reward when they choose which brand to name, and 3% of brands are collecting 71% of those recommendations¹.

What it scores: the 7 components

The Index decomposes distinctiveness into 7 measurable components, because a brand can be strong on 2 and invisible on the other 5, and the fix differs per component:

1. Lexical distinctiveness. Does the language sound like you or like the category? Measured on phrase patterns, vocabulary, and distance from competitor content.

2. Positional distinctiveness. Do you stand for something specific, or recycle the category's value props? A contested position scores; "quality and innovation" does not.

3. Evidentiary distinctiveness. Proprietary data, named methods, cases with numbers. Proof a competitor cannot photocopy.

4. Authoritative distinctiveness. Is there a named human behind the brand, with credentials and a footprint, or is it faceless?

5. Structural distinctiveness. Formats and depth the category does not publish: original research, frameworks, indices.

6. Contextual distinctiveness. Does your AI-assisted output still carry your context, or does it drift back to the average voice the model defaults to?

7. AI search visibility. Do you appear when engines answer buyer questions in your category?

Notice what the list refuses to include: follower counts, publishing volume, domain authority. Volume is how brands drown; measuring it was always measuring the wrong thing.

How the scoring works

Each component is scored 0 to 4 against a fixed rubric, evidence attached to every point: 0 means the component is effectively absent, content indistinguishable from the category default. 4 means original IP, named methodology, proprietary data, or a voice competitors cannot reproduce. A claim only counts if it is specific, supported in the text itself, and hard for a rival or a vanilla model to reproduce. "Trusted by leading companies" scores nothing. "We audited 50 category sites and 38 shared the same headline structure" scores.

The totals map to 5 bands:

The part you cannot game

Scores invite gaming, so the Index anchors on 2 tests that do not read your marketing.

The first is blind attribution: strip every logo and name from a sample of your content and your competitors', then ask a classifier to sort out whose is whose. If it performs near chance, no voice exists, whatever the brand book claims. It is the logo test, industrialised.

The second is the machine pole: generate a vanilla model draft of your exact brief and measure the semantic distance between it and what you published. 2 vanilla drafts of the same brief sit about 0.06 apart; genuinely distinct writing runs 3 to 5 times further from the machine's default. Sitting near the pole means the work added nothing the model would not have said anyway, which is why generic prompts return generic answers: the average is what the machine does unaided. You cannot prompt a classifier that has never seen your instructions, and you cannot argue with a distance.

What a low score buys you

A diagnosis, per component. Lexical collapsed but evidence is strong: the fix is voice, not more case studies. Everything strong except AI search visibility: the fix is entity and footprint, not rebranding. The Index is the input; the commercial outcome it predicts is Share of Recommendation, the share of AI answers in your category that name you.

At Ivanooo, Firoz Azees built the Brand Distinctiveness Index as the instrument inside Distinctiveness Engineering, and pointed it at Ivanooo first: our own baseline found a strong voice score stranded behind a broken entity component, which reordered our entire quarter. That is what an instrument is for. If it cannot surprise its owner, it is a brochure. See where your brand lands: get recommended by AI.