How Bristlenose reads your interviews

Bristlenose organises an interview study into a fixed set of structures. Understanding them makes the report easier to read and edit.

Sessions

Each recording is a session. Files that share a name stem form one session — see Supported files.

Participants and roles

Within a session, Bristlenose identifies who is speaking and assigns a role:

• Participants — the people being researched. Their words are the primary data. They are given codes (p1, p2, …).
• Moderators — the researchers asking questions. They are never quoted.
• Observers — silent attendees. They are excluded from quotes.

Participant codes run across the whole study: session one might be p1–p3, session two p4–p6. Moderator and observer codes (m1, o1) are numbered within each session.

The codes are the anonymisation boundary. You can add display names to help your team remember who is who, but those stay internal — exports use codes by default. See Privacy and data.

Quotes

Bristlenose extracts quotable moments from each participant. Filler words ("um", "you know") are removed; meaning, emotion, and self-corrections are kept. Where a phrase would not make sense out of context, a short clarification is added in square brackets — for example, "the [settings page]". Quotes must be at least five words.

Every quote appears in exactly one place in the report. There are no duplicates.

Sections and themes

Quotes are grouped two ways:

• Sections group quotes by the screen or task the participant was looking at. They are ordered to follow the product, like a journey.
• Themes group quotes by a pattern that recurred across participants — a shared opinion, a common workflow, a repeated point of confusion. Themes are not drawn from a fixed list; they emerge from the data. A theme needs at least two quotes; weaker observations are collected separately.

A study heavy on task walkthroughs produces mostly sections; a set of open-ended conversations produces mostly themes. Either is a valid result.