Ideas

The Eight Surfaces

The Eight Surfaces is a framework for measuring how AI assistants such as ChatGPT, Claude, Gemini and Perplexity describe a brand. It was created by Sean Stevens, founder of brand·lore. It has eight measures in two groups. Four cover visibility: Presence, Position, Coverage and Authority. Four cover perception: Sentiment, Characterization, Accuracy and Relativity.

I built it because one number hides too much. A brand can be named in most answers and still be described wrong in most of them. Eight measures show where the problem is.

Presence Position Coverage Authority Sentiment Characterization Accuracy Relativity
Fernhill Kitchen Sample brand. The values are invented.

The four visibility surfaces

These cover whether AI brings the brand up.

Presence. How often the brand is named across every answer. If I ask 12 questions of four assistants three times each, that is 144 answers. Presence is the share of them that name you.

Position. Where the brand sits when it is named. Being the assistant's pick is different from being fourth in a list of six.

Coverage. How many of the questions the brand appears on at all. A brand can be named often on two questions and never on the other ten.

Authority. How often the assistant uses the brand's own pages as a source when it explains the category.

The four perception surfaces

These cover what AI says once it has brought the brand up.

Sentiment. Whether the description is favorable, neutral or unfavorable.

Characterization. The words the assistant chooses for the brand. "Budget option" and "category leader" can both come with a favorable tone, and they place a brand very differently.

Accuracy. Whether the facts are right: what the company sells, where it operates, who runs it.

Relativity. How the brand is described next to the competitors named in the same answer.

Visibility and perception move apart

In my own runs, one model kept describing a company by the single region where it started. With search on, the same model mostly described the company as it is today. That company's Presence was fine. Its Accuracy was the problem, and no visibility number would have shown it. I wrote up those runs in How often AI looks up your brand.

How I use it

The baseline report measures all eight against your competitors, with the sources behind each answer. The monthly read tracks the ones that moved. A snapshot is smaller: two findings from one buyer question.