If old Gravity Forms survey answers disappear from an export after you rename, reorder, add or remove checkbox choices, stop editing the field while you investigate. Gravity Forms warns that these changes can alter input IDs and break the connection between stored answers and their original choices. A blank exported cell is not enough to conclude that the respondent left the question unanswered.
The warning is documented under modifying survey checkbox choices. This guide covers preserving and reconciling affected data, not changing the survey's scoring or calculating a new aggregate result.
Preserve the Evidence Before Attempting a Repair
Take a restricted backup of the form definition and relevant entries. Keep any earlier form export or backup from before the choice change. Record when the edit occurred and whether new submissions continued afterward. Avoid another round of choice reordering to see whether the answers magically return.
Choose a few entry IDs from before the change and a few after it. Prefer synthetic or explicitly approved examples. Record what the entry detail screen shows, what the CSV contains, and which field or subfield each value belongs to. Do not publish respondent names or free-text answers in a support screenshot.
Start with the standard survey export procedure to rule out a different form, missing selected fields or date filters. If the mismatch starts exactly at a field edit, the historical definition deserves closer inspection.
Build a Revision-Aware Mapping
Use a worksheet that separates the old question definition from the current one. The example below describes the columns to record; it is not a mapping that can be applied to your form.
| Mapping field |
Why it matters |
| Form and field ID |
Identifies the question rather than its label |
| Historical input ID |
Identifies the stored checkbox answer |
| Label and value at submission |
Prevents relabeling the past with today's choices |
| Revision date range |
Separates old and new submissions |
| Evidence source |
Shows whether a backup or export supports the mapping |
Keep an unresolved state when the original meaning cannot be established. Do not infer a deleted choice from its current screen position. An attractive complete chart is worse than an honest partial dataset with a documented limitation.
Preserve: Entries and form revisions. Sample: Before and after the edit. Unresolved: Never invent missing answers. Reconcile: Distinct IDs and answer states. Explanatory checklist, not a customer test result.
Recover on a Copy, Then Reconcile
Ask your developer or Gravity Forms support to compare the preserved definitions and stored data on a protected copy. If a supported repair or custom export is needed, review it against the small sample before applying it to the full set. Keep the original records intact and record each transformation separately.
Count distinct entry IDs before and after recovery. Compare answered, genuinely blank, conditionally hidden and unresolved values separately. A row-count match alone cannot prove the choices were assigned the right meaning.
Can You Replace Every Blank With the Most Common Choice?
No. That invents responses. Mark unresolved data clearly and explain how it affects any percentages or conclusions. If the historical definition is unavailable, say what cannot be recovered reliably.
For future survey revisions, preserve the old definition and plan a new question or form version when meaning changes. Test exports with a synthetic multi-choice response before reopening collection. Request help with a form-data repair using a sanitized sample and revision timeline.
References reviewed October 15, 2026. Examples are explanatory, not customer test results.