blog
On forms, response quality, and why the respondent's experience is the product.
SurveyMonkey analyzed 25,000 surveys. Open-ended questions drop completion by 10 points. The average answer is 5 words.
Surveys over 25 minutes lose 3x as many respondents. The fix isn't fewer questions — it's changing who does the retrieval.
Sprint retros are the most common recurring form in engineering. Also the worst-answered. Filled at 4:55pm Friday, from memory.
Matrix grids look clean to form creators. They exhaust respondents. Couper et al. found they triple straight-lining.
Forty years of survey research, distilled into 7 rules. Most forms break at least three.
Seven probe patterns from qualitative research, mapped to form fields you can configure today.
People lie on forms. Not maliciously — reflexively. Social desirability bias is the largest source of systematic error in org surveys.
A 95% completion rate with one-word answers is worse than 60% with specifics. Groves proved it in 2006.
Bank transfers, inventory, CRM — every data-exchange boundary automated. Forms survived. They won’t much longer.
A growing share of survey responses are AI-assisted. The question isn’t whether to allow it. It’s whether your form is designed for it.
An 85% response rate on your engagement survey isn't candor. It's obedience.
A retro filled Friday at 5pm captures two weeks through the lens of the last three hours.
NPS is the most-deployed metric in business. The number tells you almost nothing.
Closed questions give you clean data. Open-ended questions find what you missed. Pick both.