Survey exports

You can download a survey's data as Excel, Data (CSV) or SPSS, from the Download row at the top of the Questions, Cross-tab, Flows and Drop-off views on the Results tab. Every file has one row per person, holds exactly the people your results filter is showing, and comes with a codebook that explains every column. Data downloads are on paid plans.

The three downloads

Excel (.xlsx): one workbook with five sheets:
- Data (labels): the data with the question wording across the top and answers written out in words. This is the one to read.
- Data (codes): the same data with short variable names across the top and number codes in the cells. This is the one to analyze.
- Codebook: what every variable and every code means.
- Tables: every question's result as a small table (counts, percentages of the base, means), the same numbers the Questions view charts.
- Open answers: every written answer, one row each, with its question and the respondent's number.
Data (CSV): the coded data on its own, variable names across the top and codes in the cells, ready for R, Python or a spreadsheet.
SPSS (.zip): a zip of three files: the coded data (survey-data.csv), an SPSS syntax file (survey-data.sps) and the codebook (codebook.csv). Unzip them into one folder, open the syntax in SPSS and run it: it reads the data and sets every variable's label, value labels and level of measurement in one step.

Downloads follow your filter

The data downloads hold exactly the people the filter bar keeps: your Responses (live, test or both), Source, Who (completes only, or everyone who started) and Quality choices. So the default download is live completes with rejected responses left out. To include screen-outs and people who stopped part way, choose Everyone who started before you download; to include rejected responses, tick Rejected.

The written report is different: it is always written on the default view, whatever the filter says (see the survey report).

What's in each row

Each row starts with columns about the person:

respondent: their number (R1, R2 and so on, in the order people started), the same number the Questions view shows beside open answers.
response_id, status (completed, screened out or ended early, or stopped part way), end_reason (why it ended early), quality (the verdict) and quality_reasons (which checks fired).
source (survey link, panel or test link), version (which version of the survey they answered), started_at and completed_at (UTC) and minutes to complete.

Then come the answers, one or more columns per question in the order respondents meet them.

How each question is laid out

Single choice, dropdown, Likert and single image choice: one column, holding a code for the answer picked (1 for the first option, 2 for the second, and so on).
Multiple choice (and image choice with several picks): one column per option, 1 if it was ticked and 0 if the person saw the question and left it unticked.
Grid: one column per row, holding the code of the column picked on that row (1 for the first column).
Rank order: one column per item, holding the position the person gave it (1 = ranked first). An item they didn't rank is blank.
Rating: the number of stars or points given.
NPS: two columns: the score from 0 to 10, and the NPS group (1 = Detractor, 2 = Passive, 3 = Promoter).
Slider and number: the number given. A short text asked as a number holds the number too.
Date: the date, written YYYY-MM-DD.
Open text: the full answer, word for word.
Consent: 1 for Yes, 0 for No.
Screens that only show text, a picture or a video have no column.

Questions inside a loop get their own set of columns for each loop item, with the item named in the column's label, so each item's answers sit side by side on the person's row.

Blank never means zero. A blank cell means the person wasn't asked that question (routing sent them past it) or skipped an optional one. In SPSS, blanks read as missing.

The codebook

The codebook has one line per variable (its name, question number, question id, loop item, label and kind), then one line per code saying what the code means and which answer option it stands for. The codes stay the same for every respondent, whichever version of the survey they took: an answer option you removed later keeps its code, after the current ones, so older answers are never lost or relabeled.

Variable names are made safe for SPSS, R and Excel alike: letters, digits and underscores, starting with a letter, and never more than 64 characters.

By plan

Data downloads are available on the Individual, Team and Enterprise plans. On the Free plan the results are view only, and the Download row shows Upgrade to download instead. The plan that counts is the plan of the workspace the survey belongs to.