Embedded Data >1000 chars failing in New Survey Taking Experience (regression?) | Experience Community
Skip to main content
Question

Embedded Data >1000 chars failing in New Survey Taking Experience (regression?)

  • January 29, 2026
  • 0 replies
  • 3 views

Forum|alt.badge.img

We’re seeing a breaking change in Qualtrics’ New Survey Taking Experience related to embedded data length limits and are looking for guidance or confirmation from Qualtrics staff or others who’ve hit this.

Use case
We deliver JavaScript/TypeScript-based cognitive assessments via Qualtrics. Each assessment produces JSON (trial-level + scoring data) that we store in embedded data fields. These values commonly exceed 1,000 characters but remain well under the documented 20KB embedded data limit.

Documentation
Qualtrics docs state:

“Each value of an embedded data field should never exceed 20KB (20,000 bytes).”
https://www.qualtrics.com/support/survey-platform/survey-module/survey-flow/standard-elements/embedded-data/

Observed issue
In the New Survey Taking Experience, survey submission fails when embedded data values exceed ~1,000 characters. At submit time, Qualtrics throws a schema validation error such as:

badSchema: body.ed.<FIELD_NAME> does not meet maximum length of 1000

The assessment itself runs correctly; only submission fails.

Legacy experience comparison
The same survey works in the legacy survey taking experience. We have confirmed successful legacy submissions with embedded JSON values ~1,500 characters long.

Reproduction

Why common workarounds don’t work

  • Compressing or transforming the JSON would break established downstream data pipelines.

  • Splitting data across multiple embedded fields isn’t viable: sessions may include 10–20 tasks with highly variable output size, and total size is already within documented limits.

Impact
This appears to be a regression or undocumented constraint in the new survey experience. It blocks multiple ongoing NIH-funded research projects that rely on Qualtrics’ documented embedded data behavior.

Question
Is the ~1,000-character limit in the new survey taking experience:

  • expected behavior,

  • a known regression, or

  • something that can be configured or fixed?

Any official guidance or confirmation would be greatly appreciated.