Create intermediate Works with: Any AI

Open-Ended Response Coding Guide

Table output Has example

Turn comments into themes.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Code open-ended survey responses. Question asked: [QUESTION]. Responses: [RESPONSES]. Research goal: [GOAL].

Process: read all responses; develop 6 to 10 themes inductively with definitions and inclusion rules; code each response (allow multiple themes); compute theme counts and shares of respondents (count / total responses, shown); pick representative quotes (exact wording, anonymized); note outliers and contradictions; and flag responses that need follow-up (requests for contact, safety concerns).

Return a codebook table: Theme | Definition | Include if | Exclude if | Count | Share | Example quote, and a coded response table for the first 20 responses. Describe a reliability check (a second coder on a sample, simple agreement percentage = agreements / total, noting chance-corrected measures exist). Summarize the main insights in 150 words and mark uncertain codes.

Show the coding of the first five responses as a worked example, say how to treat a response that fits two themes, and set a rule for the 'other' category (review if it exceeds 10 percent of responses).

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

Question: "What is the one thing we should improve?" 80 responses.
Codebook: Speed (slow loading, delays) - include if mentions waiting; exclude general "make it faster" about work tasks. Pricing; Support; Missing features; Navigation; Other.
Counts: Speed 22 (27.5 percent); Missing features 18 (22.5); Pricing 12 (15); Support 9 (11.3); Navigation 8 (10); Other 11 (13.7). Check: 22+18+12+9+8+11 = 80.
Example quote (Speed): "The report takes a minute to open every time."
Multi-theme: a response mentioning pricing and support is coded twice.
Reliability: second coder on 20 responses; agreement 17 of 20 = 85 percent.
Insight (150 words): speed and missing features account for half of requests. Mark uncertain codes.

• response-coding