CASE 04 · RESEARCH AUTOMATION · REAL PROJECT
Automated Verbatim Quality Checker.
A pattern-detection system I built to flag gibberish, non-answers, and copy-pasted text in open-ended survey responses. Cutting manual QC time by over 60%.
THE CHALLENGE
Open-ended survey verbatims needed slow, inconsistent manual review to catch gibberish, blank answers, and text copy-pasted from AI tools or search engines.
MY ROLE
As the senior project manager overseeing research delivery, I identified this as a recurring bottleneck across multiple projects and took ownership of designing and building a solution myself, rather than waiting on a dedicated development resource.
WHAT I DID
Built a configurable pattern-detection system in Excel using VBA and regex, able to flag gibberish and keyboard-mash text, blank or placeholder answers, repeated-character spam, and signs of copied or AI-generated content such as unusual length or list-like formatting. Detection rules are adjustable per project, so the logic adapts to each client’s quality criteria rather than being a fixed, one-size-fits-all filter.
HOW IT FLOWS
- STEP 1Raw verbatims collected
- STEP 2Pattern detection rules applied
- STEP 3Entries flagged for review
- STEP 4Clean dataset passed to analysts
THE OUTCOME
Reduced manual verbatim-checking time by over 60%, freeing analyst capacity for higher-value interpretation work instead of manual data cleaning, and improved consistency in what counted as a flagged response across projects and analysts.
SEE IT IN ACTION
How it works: this demo reproduces the detection logic at a simplified level. The production version (built in Excel VBA using configurable regex rules) runs across full survey datasets, flags entries for review before analysts see them, and lets the rule set be adjusted per project. Categories shown: gibberish/keyboard-mash patterns, non-substantive answers, blank/placeholder responses, repeated-character spam, and signs of copy-pasted or AI-generated text (excessive length, list-like formatting).
LET’S START WITH A CONVERSATION