To export individual Gravity Forms survey responses, use Forms > Import/Export > Export Entries, select the survey form, choose the fields and filters, and download the CSV. The Survey Results screen is an aggregate view. Gravity Forms does not provide a direct export of that summary; export the underlying entries and rebuild the summary in your analysis tool.
The official entry export instructions describe field selection and filters. The Survey Results guide explains the reporting view. The steps below help you check that the exported data answers the same question as your report.
Define the Population Before Clicking Export
Write a one-sentence definition, such as: completed responses to form 18 during the selected date range, excluding the test submissions we can identify. Then record the actual filters available in your installation. Do not assume the results screen, Entries list, and export screen all have the same filters selected.
Use entry IDs to distinguish records. Names and email addresses are poor reconciliation keys and often are not needed for analysis. Keep respondent details out of screenshots and reusable examples. An analyst who only needs response counts should not receive a file containing unnecessary personal information.
Make a Small Export First
Choose a short date range containing a few known responses. Include the entry identifier and the survey questions you need. Keep the raw download unchanged and work from a copy. Record the form, selected columns, filters, export time, and row count next to the file.
Open a small sample as raw text before importing into a spreadsheet. Check commas inside answers, multiline responses, non-English characters, and identifier columns with leading zeros. Then explicitly choose text types where automatic conversion would change the data. Do not enable spreadsheet execution for formula-like respondent input.
Illustrative diagnostic aid. Record your own results; no customer testing is implied.
Map Questions Without Erasing Their History
Build a mapping sheet that connects the form ID, field or subfield ID, current label, and exported column. Where a question has choices or scores, document what the exported value represents. A displayed label, a stored choice value, and a numeric score are different concepts.
If the survey changed while responses were being collected, find the relevant historical definition before relabeling older answers. Do not silently assign today's choice score to an older response. Use the historical choice and score review when this distinction affects the analysis.
Reconcile Counts at the Right Level
Count distinct entry IDs before expanding multi-select answers. One response with three checked choices should not accidentally become three respondents in the headline count. Separately record blank answers, questions hidden by conditional logic, and responses excluded by your filters.
| Check |
Question to answer |
| Respondent count |
How many distinct eligible entry IDs are in the export? |
| Question denominator |
How many of those entries contain an answer to this question? |
| Multi-select total |
Are you counting respondents or selected choices? |
| Date boundary |
Did both views use the same range and timezone interpretation? |
| Form revision |
Did labels, values, or scoring change during collection? |
If the CSV and dashboard disagree, compare the missing or extra IDs before adjusting formulas. A blanket percentage correction can make one chart look right while leaving the underlying dataset wrong.
Worked Example: Respondents Are Not Choice Totals
Suppose a synthetic export contains five distinct entry IDs. Four respondents answered a multi-select question and one left it blank. If two respondents selected two choices and two selected one choice, the result is five respondents, four people answering this question and six selected choices. None of those counts is interchangeable.
If you report the share of people who answered the question, use four as the denominator and state that choice explicitly. For a share of all eligible respondents, use five instead. Keep the blank response in your reconciliation rather than deleting it to make a chart match. The six selections describe answer choices, not six people.
Deliver an Analysis File, Not an Unexplained Download
Include the report definition, field mapping, exclusions, and a count reconciliation with the analyst's copy. Keep the raw original accessible only to authorized staff and apply the organization's retention policy. If uploads appear in the export, review the file-download URL exposure checklist before sharing it outside the team.
This process does not claim that every installation exports every question type identically. Test the types, add-ons, and versions you actually use. HandL WP can help trace a survey export discrepancy using a small redacted example, the form definition, and the filters that produced it.
Count-reconciliation example and introduction updated October 7, 2026. Examples and diagrams are explanatory, not measurements from a customer site.
If the discrepancy appears only in command-line output, check the CLI display layer: Gravity Forms CLI Shows Array Values Incorrectly: Check CLI 1.9.1.