Run your codebook
on every transcript.

A different approach to a priori coding.

A priori coding is the practice of defining your categories before you touch the data: a fixed codebook built from a theory, a research question, or a client’s briefing, then applied consistently across every interview, focus group, or open-ended response in a study. It sits opposite emergent, or a posteriori, coding, where categories are built up from what respondents actually said as you read. Most real studies use both, but a priori coding is what makes results comparable across waves, markets, and moderators.

Where the manual process breaks down

The method is simple to describe and slow to run by hand. A researcher reads a transcript, holds ten or twenty category definitions in their head, and tags each passage while flipping back to the codebook to check a borderline case. Two coders on the same project drift apart within a week unless someone runs a formal inter-rater reliability check. By the time forty interviews are coded, the categories that mattered in week one have quietly shifted, and nobody wants to recode the early batch.

How Speak AI applies your codebook

Speak AI reads every interview on three layers: the words a participant actually said, the tone, emotion, and energy in how they said it, and, where you upload video, the visual cues in the room. Your a priori codebook, categories, definitions, and decision rules, is loaded once and applied to every transcript the same way, every time, with the source passage and a confidence score attached to each tag. Code frequency and coverage trend over time as new interviews come in, and instrumentbord du kan tilpasse og white-label track how themes shift wave over wave instead of living in a static spreadsheet. Ask the same question across the whole study through MCP-connected AI chat, with your choice of Claude, ChatGPT, or Gemini underneath.

What a priori coding looks like in practice

Most a priori projects follow the same shape, whichever tool runs it:

• Establish the codebook from your research questions or a client brief, with a clear definition for each category.
• Apply the codes to every transcript the same way, instead of re-deciding borderline cases interview by interview.
• Compare code frequency and coverage across segments, waves, or markets.
• Flag passages that don’t fit and log where the framework needs an emergent code added.

From a static codebook to comparable results

The result is a codebook that behaves the same on interview one and interview two hundred, with a full audit trail back to the original recording for every tag. That consistency is what a client-facing report or a dissertation defense actually needs: not just codes, but codes applied the same way every time. One global market research firm ran its a priori framework through this workflow across a multi-country study and saved $60K and 950 hours without changing its codebook or its analysts. If you’re deciding between a rigid framework and letting themes emerge, it is worth comparing the other qualitative coding approaches or looking at real coding examples before you lock a codebook in.