Survey open-ended answers

Code the free-text answers of a survey into themes and sentiment, and pull out the ones that name a specific change.

Run this on your file Download the template CSV

Before you run it: Replace the theme options with the themes of your own codebook. You can edit every question in the runner.

Columns to upload

Name the columns in your CSV like this. Extra columns are fine: every column of the row is sent with the questions.

Column	What it holds
`respondent_id`	Respondent ID
`question`	The survey question
`answer`	The respondent's free-text answer

Questions

Each question adds a column to your file. The answer is one of the listed values, or Unsure when the model's confidence is below the cutoff: 70% for pick-one and scale questions, 80% for yes or no.

New column	Question	Possible answers
`theme`Pick one	Which theme is the answer mainly about?	price Cost, value for money, discounts quality How good the product is ease of use How easy the product or service is to use support Help from staff or customer service delivery Shipping, speed, packaging flexibility Skipping, pausing, changing or cancelling an order or plan other None of the themes above
`sentiment`Scale	How does the respondent feel?	very negative negative neutral positive very positive Lowest to highest
`actionable`Yes or no	Does the answer name a specific change the company could make?	yes It names a specific change no It names no specific change
`off_topic`Yes or no	Is the answer unrelated to the question, empty of meaning, or a joke?	yes It does not answer the question no It answers the question

Example row

Column	Value
`respondent_id`	R-1042
`question`	What is one thing we could do better?
`answer`	Let me skip a month without calling. The wine is great but I travel a lot.

Answers from a test run on this row

Column	Answer	Confidence
`theme`	flexibility	100%
`sentiment`	Unsure	42%
`actionable`	yes	87%
`off_topic`	no	81%

More research use cases

Use case	What it does	New columns
User interview snippets	Tag quotes from user interviews by the pain they describe and how serious it is.	`pain_point` `severity` `workaround` `feature_request`
Literature review abstract screening	Screen the titles and abstracts of a literature search against your inclusion criteria before you read the full texts.	`design` `population` `outcome` `include`
News and media content analysis	Code news articles by frame and tone for a media study.	`frame` `tone` `cites_study`
App store reviews	Sort app reviews into bugs, requests and praise, and see which version the problems come from.	`category` `sentiment` `crash`
Social listening posts	Code public posts about a topic by stance, and separate first-hand experience from repeated claims.	`on_topic` `stance` `first_hand` `health_claim`