Skip to main content
The Export button on the Summary and Responses tabs opens one modal for every export. It always opens with the default settings and does not remember what you chose last time. There are two kinds of export: The Standard export is the file Formbricks has always produced, with one change: the Excel sheet is now named Response Data instead of Sheet1. A pipeline that reads the Excel file by sheet name must use the new name, and one that reads the first sheet or a CSV file is not affected. It is plain UTF-8 text without a byte-order mark and with \n line breaks. A question a respondent was not shown is an empty cell, exactly like an unanswered one.
Anyone who can read a survey can export it, including read-only seats.

Choose what to export

1

Filter the responses

Set the filter and time range on the Summary or Responses tab first. The export contains the responses the filter matches. To export only some responses, tick their rows in the Responses table and select Download in the selection bar: it opens the same modal for the selection.
2

Open the Export modal

Select Export. The Export options modal opens with the settings of the Standard export. A count box at the top says how many responses are included, for example “Include 1,240 responses”, and below it what the number is based on, such as your filters, the Low quality responses left out and the total: “Matching your current filters · 3 Low quality left out · 1,300 responses total”. The Export button waits until the count has loaded.
3

Pick the options

  • Format: .csv or .xlsx.
  • Add SPSS syntax file (.sps): CSV only. It adds a syntax file that reads the CSV into SPSS. It needs answer codes, so it switches Text only off.
  • Display questions and Display answers: Text only, Code only or Text and code, one choice for the column headers and one for the cells.
  • Include columns: tick the columns the file keeps, or Select all. A multi-select question written as one column per option lists its options and its Other text under it, so you can leave one out without the rest.
  • Include Low quality responses (N): shown when the responses you export contain Low quality ones. It is ticked for the Standard export and unticked once you choose any Analysis option, until you change it yourself. If your filter selects Low quality responses, the filter decides and the checkbox is not shown.
  • Advanced options:
    • Split multi-select into 1/0 columns: one column per option of a multi-select. It is ticked once you choose any Analysis option, and always on with the SPSS syntax file. Untick it to write each multi-select in one cell instead. Ticking it on the Standard settings makes the export an Analysis export.
    • Add datamap and Add unmapped values, both for CSV and Excel. On CSV, each one adds a file, so the download becomes a zip.
    • Compress into .zip downloads a single CSV as a zip.
4

Export

Select Export. The modal shows the progress, for example “Exporting… 600 of 1,240 responses”, and you can close it: the download starts when the file is ready. A toast confirms it. If it fails, the toast offers Try again.
The export runs in the app and streams the file as it is built. You can move anywhere in the app while it runs, the survey editor included. If you reload or close the tab, the browser asks you to confirm first, and the export stops if you do. Only one export runs at a time in your browser, and an Excel file is built by one export at a time on the server, so a second Excel export waits its turn.

Attachments

If the survey has a file upload question, the modal also has an Attachments section with a Download attachments button. It downloads the files respondents uploaded, in one zip, for the responses the current filter matches, then closes the modal. The attachments section is not offered when you open the modal for ticked rows.

Files in an Analysis export

An export that has more than one file is always a zip. The datamap and the unmapped values list are only in the file when you tick them.

Size limit for Excel

An Analysis export as Excel is built whole in memory, so there is a limit: 1,000,000 cells, counted over the data sheets. A cell is one position in the sheet, so the count is the number of columns times the number of responses plus the header row, blank cells included. Text and code has two data sheets and doubles the count. The Standard export as Excel has no such limit. As a guide, with Text only or Code only: With Text and code, halve those numbers. The columns are one per question, one per option of a multi-select, one per matrix row, and the response columns such as timestamp, tags and metadata. When a selection is over the limit, the Excel option is greyed out, the modal says “Excel is off for this export: the file would hold N cells, more than the 1,000,000 an Excel file can”, and Export stays off. A button switches the format for you. If your settings produce a zip, it reads Export as a zip of CSV files, which we recommend. Otherwise it reads Export as CSV. CSV and zip exports have no size limit. You can also narrow the filter until the selection fits.
Every file of an Analysis export is UTF-8, so Arabic and other non-Latin text stays readable. CSV files start with a byte-order mark so that Excel and SPSS detect the encoding.

What the columns hold

  • The first columns are response_id, language and outcome. The other response columns follow if you included them.
  • Each question has a column named by its question code. A matrix has one column per row, named <code>_<row code>.
  • A single-select or picture selection question has one cell with the answer code. If the question has an Other option, the text typed into it goes to <code>_text.
  • A multi-select question has one column per option, named <code>_<answer code>. A cell is 1 when the option was chosen, 0 when the question was answered without it, and blank when the question was not answered or not shown, or when show if did not offer the option to the respondent. A multi-select picture selection follows the same rule. Responses from before Formbricks 6.2 don’t record what was not shown and keep 0. Options added after the survey was published also get 0 for earlier responses that answered the question; the datamap notes this.
  • With Split multi-select into 1/0 columns unticked, a multi-select question is one column named by its question code. The cell lists the chosen answer codes in option order, separated by ; , for example 1; 4. In the text data it lists the answers as given, and an Other answer as its text. Its Other text keeps its own <code>_<answer code>_text column.
  • A ranking question has one column per option, holding the rank the respondent gave it (1 is first), blank if the option was not ranked.
  • Rating, NPS, CSAT and CES questions hold the number. Number inputs are exported as numbers.
  • Address, contact info, file upload, date, consent and calendar questions hold the same cell as in the Standard export, under the question code.
  • Calls to action and welcome cards are left out of an Analysis export.
  • A question the respondent was not shown (a show-if kept it from them) is blank in every one of its columns, never 0. So is a matrix cell whose row or column was not shown. Responses collected before “not shown” was recorded keep the rules above.
  • Blank always means “not asked or not answered”. There are no user-missing values. If you need codes such as 98 or 99, add them as ordinary answer options.
  • With Text and code, the header cell in the Labels sheet and labels.csv adds the question text: q2; How satisfied are you?. The column name stays q2.
  • In CSV files, a line break inside a cell becomes a space, because SPSS cannot read records that span lines. The Excel file keeps the line breaks.
  • Code cells are written as text in Excel, so codes such as 01 and 98 keep their form.

Datamap

The datamap explains the export. Tick Add datamap to get it. It is in long format, with one row per column and per answer code, and one row for a column that has no codes. The headers are always in English, in this order: Embedded Data and response information columns are listed with their types. A legacy Embedded Data field that is named like an export column gets an ed_ prefix and a note.

Unmapped values

Answers are stored as the text the respondent saw, so Formbricks finds the code of an answer by matching that text with the options of your survey, first in the language of the response and then in any language. Tick Add unmapped values to get the list. An answer that matches no option is an unmapped value. It is never guessed and never counted as Other. In the data, its code cell holds UNMAPPED: followed by the text as given. The cell is marked even when you do not add the list. The Unmapped values sheet, or unmapped-values.csv, lists each one: Typical causes are an option that was renamed or deleted before the survey was published and locked, or an answer sent through the API. If a question has an Other option, text that matches no other option reads as an Other answer and is not listed. A multi-select answer that matches no option has no cell, so it appears only in the unmapped values list.

SPSS syntax file

With Add SPSS syntax file the zip contains labels.sps, UTF-8 with a byte-order mark. It reads the CSV of codes into SPSS (data.csv, or codes.csv with Text and code) and sets up the labels, so that you do not have to. Before you run it:
  • Open the file in SPSS. Near the top is a line CD 'C:\path\to\the\unzipped\export'. It is a placeholder: set it to the folder where you unzipped the export, then run all.
What it does:
  • Reads the file as UTF-8 (/ENCODING='UTF8'), skipping the header row.
  • Reads the data with an explicit variable list, so SPSS never guesses a type from the first rows. A column is numeric only when every answer code is a plain integer. If an answer code is a letter, the column is a text variable, which SPSS can label but not compute with.
  • Sets variable labels (at most 256 bytes) and value labels (at most 120 bytes), cut on a character boundary. Quotes are doubled and line breaks become spaces.
  • Defines a multiple response set for each multi-select with at least two option columns.
  • Does not define user-missing values. Blank is system-missing.
  • Reads a column that holds UNMAPPED: cells as text, keeps the original in a <column>_raw variable, then converts the column with ALTER TYPE. SPSS reports a warning for every UNMAPPED: cell it cannot turn into a number, and the file raises the warning limit so the run does not stop early.
  • Reads the timestamp as text, for example 2025-05-06 14:30:00 UTC in the display time zone of your organization, and adds a date-time variable timestamp_dt made from its first 19 characters.

Low quality responses

Mark a response as Low quality when it should not count, for example a speeder, a duplicate or a test. Nothing is deleted, and you can restore the response at any time.
  • On a response card, use Mark as low quality in the header, and Restore as complete to undo it.
  • In the Responses table, tick several rows and use the bulk action in the selection bar.
  • The response status dropdown in the filter, under “Show all responses that match”, has Low quality responses, so you can review them. The other choices do not hide them.
  • Read-only seats see the outcome of each response but do not get these buttons.
  • Through the API, set outcome to lowQuality, or to null to restore, with PATCH /api/v3/responses/{responseId}. POST /api/v3/responses/batch-outcome does it for up to 100 responses at once.
What changes for a Low quality response:
  • It is left out of the Analysis export by default, and the count line says how many were left out. Tick Include Low quality responses to keep them. This also holds when you export ticked rows. If your filter selects Low quality responses, the export contains them.
  • The Status column of the Responses table and the response card show a Low quality badge. The Analysis export has an outcome column that reads lowQuality.
  • Its Feedback Records are removed, so it drops out of Dashboards. Restoring the response creates them again.
What does not change: the Summary, response counts, API lists, webhooks and integrations keep counting Low quality responses. Webhooks and API responses carry outcome, so you can filter on it yourself. The Standard export includes Low quality responses too, unless you untick Include Low quality responses.
Whoever marks responses needs edit rights on the survey.