Analysing
Get the data out, in a shape you can use.
Responses export as CSV, with one column per matrix row and a separate column holding the free text behind “Other (please specify)”. Any single response can also be turned into a PDF using the browser's own print-to-PDF — no PDF library is shipped to respondents, so the survey itself stays light.
Two export routes for the data you will analyse, both boring on purpose, and one signed route out to other systems. Boring exports are the ones that still work in three years.
The CSV: one column per matrix row
Matrix questions are where most survey exports go wrong. A matrix asks several statements against one shared scale — “rate each of these on agree to disagree” — and a naïve export flattens all of it into a single cell of comma-separated mush that you then spend an hour splitting apart.
Here, each row of the matrix becomes its own column. An eight-statement agreement grid exports as eight columns, one per statement, each holding that respondent's point on the scale. You can pivot it, chart it or run a mean across the columns without touching the data first. If you are planning a comparison between items, ask them as one matrix rather than as eight separate questions — see the full list of question types for what a matrix can and cannot do.
“Other” text keeps its own column
When a respondent picks “Other (please specify)” and types something, two facts exist: that they chose Other, and what they wrote. The export keeps both, in separate columns. The choice column stays clean and countable — you can total how many people chose Other without stripping free text out of the same field — and the detail column holds the wording exactly as typed.
The free text is stored separately in the product as well, not glued on to the option label. It is also kept only while the question is visible and the option is still selected: if a respondent changes their mind, or conditional logic hides the question entirely, the stray text does not linger in the response and will not appear in your export.
What ends up in a row
| What you asked | How it exports |
|---|---|
| Matrix | One column per row of the matrix. |
| Choice with “Other” | The chosen label in one column, the typed detail in another. |
| File upload | The file itself stays in the private bucket — a CSV cannot carry a document. Files are reached from the response inside the app. |
| Hidden fields | Stored with the response, so the source or cohort you passed in the query string travels with the answers. |
A PDF of one response
Individual responses are browsable one at a time, and any one of them can be printed to PDF through the browser. That is genuinely all it is: a page laid out to print sensibly, and Ctrl/Cmd + P → Save as PDF.
The reason to do it this way rather than generate PDFs server-side is weight. Client-side PDF libraries are large, and shipping one would mean every respondent downloads code that only staff ever use. Print-to-PDF costs the respondent nothing, works on every modern browser, and produces a file the recipient can open without asking what it is.
A per-response PDF is the right artefact for casework — attaching one application to a file, sending a single submission to a colleague, keeping a record of what someone actually agreed to. It is the wrong tool for analysis. For that, use the CSV.
Clear your test responses before you export
Every survey gets tested, and every test leaves rubbish behind. “Clear test responses” removes them along with their uploads and any partially saved sessions, so your export starts at the first real respondent. Run it immediately before you publish the live link — separating real from fake in a spreadsheet afterwards is far more tedious than it sounds.
The third route out, and what does not exist
There is a third route, and it is deliberately narrow. Every response
can be posted to an https URL of your choosing, signed with
HMAC-SHA256 using a per-webhook secret so the receiver can prove the
message came from us, with a delivery log; the detail is on the
webhooks page. Email notifications can
also tell your team a response has arrived — the answers themselves are
not in that message.
There is no Zapier app, no Google Sheets connector, and no API-driven sync into a warehouse. That is a deliberate decision rather than a gap waiting to be filled: integrations multiply the number of places your respondents' answers end up living, and the whole point of storing survey content and responses in London is knowing exactly where they are. For analysis, data leaves through a CSV you downloaded or a PDF you printed, both of which you can account for.
Common questions
- What format is the export?
- CSV. It opens directly in Excel, Numbers, Google Sheets, R, Python or SPSS without conversion.
- How are matrix questions exported?
- Each row of a matrix becomes its own column in the CSV, so an eight-statement grid produces eight columns rather than one combined cell.
- Where does “Other (please specify)” text go?
- Into a separate column from the option label. The label records that “Other” was chosen; the adjacent column holds what the respondent typed.
- Can I export a single response as a PDF?
- Yes. An individual response is laid out for printing and you produce the PDF with your browser’s print-to-PDF. No PDF library is shipped to respondents.
- Are there webhooks or a Zapier integration?
- Webhooks are. Each response can be posted to an https URL, signed with HMAC-SHA256 using a per-webhook secret, with a delivery log. There is no Zapier app and no Google Sheets connector. For data you will analyse, CSV export and the browser’s print-to-PDF are still the two routes.
If you want to see what the numbers look like before you download anything, the live results view charts every question as responses arrive. For the wider picture of how building, collecting and analysing fit together, start at the product overview.
The export starts in the builder.
Ask eight statements as one matrix and you get eight columns you can pivot straight away; ask them as eight separate questions and you will be tidying up afterwards. Decide the shape now, while the survey is still empty.
- One CSV column per matrix row
- “Other” free text kept in its own column
- Any single response printed to PDF