Get your data out, in the format your analysis needs
Six formats, on every plan including Free. The distinction that matters is not the extension — it is whether the file still carries your codebook, and only two of them can.
What you can write, and what survives
| Format | File | Keeps labels | What to know |
|---|---|---|---|
| SPSS | .sav | script only | The one-click download is unlabelled; a labelled file with value labels, typed missing values and measure levels comes from an analysis script. Variable names truncate at 64 characters. |
| Stata | .dta | script only | Reachable through the export API rather than the download menu today. Names truncate at 32 characters, and a name starting with a digit is prefixed. |
| CSV | .csv | no | The lowest common denominator: no labels, no types, opens anywhere. |
| Excel | .xlsx | no | One sheet of values. Timezone-aware timestamps are rejected by the writer, so they are stripped to naive values first. |
| Parquet | .parquet | no | Columnar and typed — the only format here that preserves timezones. The right choice for a warehouse or a data-lake landing zone. |
| SQLite | .sqlite | no | A single table named data, with datetimes stored as ISO text. Handy for a self-contained copy you can query. |
“Script only” means the metadata is written by the engine's writer from an analysis script, not by the download button.
Labels are the whole question
A survey dataset without its codebook is a table of integers, and rebuilding one after the fact is how a refusal ends up counted as a category. Because the questionnaire is code, each variable already declares its label, its value labels and which codes are missing and why — so an export can carry that through instead of throwing it away.
That is the argument the reproducible survey research hub makes at length, and the SPSS page shows in one picture.
Questions
Do exports cost anything?
No. Every format here is available on every plan, including Free — getting your own data out is not something we gate. Connectors, which push data into an external system on your behalf, are the part that needs Plus or Pro.
Is there a row limit?
100,000 rows per export. Beyond that, use a connector into a warehouse or database, or write the file from an analysis script.
Which format keeps the codebook?
SPSS and Stata, and only when written from an analysis script through the engine's writers. The one-click downloads carry data and variable names, not labels.
Can I export JSON?
Not as a file format — the export endpoint accepts CSV, Parquet, Excel, SPSS, Stata and SQLite. If you want JSON over the wire, the custom HTTP connector posts a JSON body to an endpoint you control.
See it on a real study
The demo is a complete study — questionnaire, responses, analysis — with no account and nothing to install.