Data Quality: how it works.

Every survey that runs through Glow's Data Quality tool is screened by more than a dozen automated checks, graded A to F, and returned with a clear record of what was removed and why. This article is the map; the other articles in this category cover each screen in detail.

Mostly automated, one manual check

The tool is roughly 90% automated. Twenty-one checks run on every response the moment a survey is loaded: traps, speeders and skimmers, bad open ends, straightliners, duplicates, pattern checks, timing checks and returning panelist checks. The one manual job for Ops is verifying the open-ended answers on the OE Checks tab. Once those are verified, every risk score in the tool can be trusted at face value.

The workflow

Load the survey. Upload the data file and questionnaire. Every automated check runs on its own.
Run the AI checks. On the OE Checks tab, run AI relevance and AI written across every text question before reading anything yourself.
Verify the open ends. Scan the flag groups, correct anything the AI got wrong, and reflag where needed.
Review Respondent Risk. Filter, inspect the evidence behind each flag, and decide your deletion line.
Delete and report. Delete the bad responses, remove them in Glow under Capture > Responses > Batch delete, then share the Client Report and the Panel Partner Export.

When is a survey done?

The OE flags are verified.
No high-risk respondents remain kept: a finished survey should show 0% high risk remaining.
The deleted IDs are batch-deleted in Glow.
Both reports are sent.