Where guides live

Inside any project, the Interview Guides tab lists every guide you’ve created. Each card shows a status badge — Draft, Approved, or Archived — plus interview type, audience type, and version. Only approved guides can drive auto-interviews.

Creating a new guide

Click Create new interview guide. You’ll pick an interview type, then choose how to build the guide: Auto-generate (Candor writes it from your learning goals) or Create your own (you write the questions). Either way you finish by reviewing and approving.

Step 1: Choose an interview type

Four options, each shaping how the guide gets generated and how synthesis later interprets the data:

Problem discovery — exploring unknown problems in a market. Open-ended sections, no hypotheses to validate.
Problem validation — testing specific problem hypotheses. The guide will include a card-sort section if you give it five or more hypotheses (more on that below).
Concept testing — evaluating a specific product concept. Sections probe what’s working, what’s confusing, and who the concept is for.
Price testing — exploring value perception, willingness-to-pay anchors, and price sensitivity.

Step 2: Auto-generate or create your own

Pick how you want to build the guide:

Auto-generate: you write learning goals and Candor drafts a full, methodology-backed guide. Best when you want depth and structured synthesis.
Create your own: you write the exact questions you want asked. Best when you have a specific script in mind. Covered in “Writing your own guide” below.

Auto-generate: write learning goals

Each learning goal is one specific thing you want to understand from this set of interviews. Click Add learning goal and type the goal as a short, concrete sentence. The order you list them in becomes the priority. Three to seven goals is a reasonable range; more than that and the guide gets thin coverage on each.

For problem validation, you’ll also enter explicit problem hypotheses. These are statements that should be either confirmed or contested by the interview evidence.

Auto-generate: generate the guide

Click Generate guide. Candor takes a few minutes to draft a complete guide: sections with intent, question variants, interviewer notes, re-anchor variants for personas you’ve already interviewed, and stopping criteria per section. You’ll be redirected to the guide detail page when it’s ready.

Writing your own guide

Choose Create your own after picking the interview type to write the exact questions you want asked. The interviewer asks them in order, as written.

Write your questions

Put one question per line in the text box. Number them however you like (Q1:, 1., or just plain lines) or paste them from a doc. You can write as few as a handful of questions or a full script (up to 30).

Allow AI follow-up questions

A toggle controls whether the interviewer can go beyond your questions. Turn it on and the interviewer may ask follow-up probes based on what the persona says, then move on. Turn it off and the interviewer asks only your questions, word for word, and nothing else. On produces richer interviews; off gives you exact control.

Analyze guide (optional)

Click Analyze guide for a quick check against the same quality bar Candor uses for generated guides: leading questions, confusing or double-barrelled questions, hypotheticals, and whether the guide fits the interview type. Each issue includes an explanation, and where there’s a clean one-for-one fix, an Update question button rewrites that question in place. The check is advisory: fix what you want and proceed, or ignore it.

Concept and price testing

For concept and price testing, each concept you supplied is shown to the persona automatically before your questions run, so you can ask about them directly (for example, “Which is most interesting to you?”). One thing to know: a hand-written guide produces lighter synthesis for concept and price testing than an auto-generated one. Auto-generated guides build the structured per-concept sections that synthesis uses for comprehension, per-concept verdicts, and willingness-to-pay analysis; a manual guide’s report is built from the raw interview signal instead.

Review goals and approve

Click Next and Candor derives learning goals from your questions so synthesis knows what the interview is meant to learn. Review and edit them, then Approve to lock the guide in for interviews.

Reviewing a draft

The guide detail page lays out what you got. Read through it before approving:

Learning goals — numbered list with the priority chip you set.
Sections — each section has a title, a purpose statement, interviewer notes (instructions to the auto-interviewer), and one or more questions with intent. Section types include Core (always run), Standard (usually run), and Conditional (run only when triggered, e.g. top-ranked hypotheses in problem validation).
Front-end re-anchors — two question variants the interviewer uses to open: one for fresh personas, one for personas you’ve interviewed before. The latter references prior sessions to maintain continuity.

Card-sort mode (problem validation only)

When you have five or more problem hypotheses, the generator switches to a card-sort methodology. Personas sort the hypotheses into four pain buckets (very challenging, somewhat challenging, not challenging, not applicable), then the interviewer probes the top-ranked ones in depth. The guide detail page surfaces a banner explaining when card-sort is active.

Vocabulary warnings

Draft guides may show an amber warning banner listing any phrases that are too jargon-heavy, too vague, or likely to confuse a persona. Each flag includes the offending statement and a suggested rewrite. Edit the inputs, regenerate, and the warning clears.

Approving a guide

Once a draft looks right, approve it. Approved guides are immutable — they get assigned a version number, and any future edits create a new version (the old one stays available for reference). Drafts can be deleted; approved guides can be archived but not deleted.

Building an interview guide