Quick answer: Abandonment is about perceived effort, not question count. Three fixes move completion the most: hide whatever does not apply to this respondent with conditional logic, let an interrupted form be resumed from the same device without making anyone create an account, and keep the layout honest on a phone. A 25-item instrument that only shows 15 relevant questions and survives an interruption will out-collect a 10-item form that restarts from zero.
The Three Ways a Form Dies Halfway
- Irrelevance. Matrix grids the respondent has no opinion on feel like homework, and the length is visible before the value is.
- Interruption. Phone rings, tab closes, someone walks away mid-cohort. If nothing is kept, the answer is not late โ it is gone.
- Layout drift on mobile. Two-column scales that collapse badly, or buttons that end up below the fold, cost you the people who answer on a commute.
Resuming Should Not Require an Account
The usual fix โ "sign in to save your progress" โ adds the friction you were trying to remove, and it is a non-starter for anonymous feedback. A better pattern is to keep the draft on the respondent's own device and ask, when they return, whether they want to pick up where they stopped. That is what FormLM's fill pages do: the draft is keyed to the application rather than to a share link, so a re-published form or a fresh link does not orphan the in-progress answers; pre-filled values that came from URL parameters are excluded from the draft, because those were handed to the respondent, not typed by them; and the restore prompt is a question, not an automatic overwrite โ answering "not now" leaves the fresh form alone.
A draft is not a submission either. Nothing enters the dataset, and no report is generated, until the respondent actually submits โ so nobody's half-finished answer skews a cohort.
Cut Perceived Length, Not Content
Conditional logic is the cheapest completion-rate lever available, and it belongs in the free plan for that reason: show a follow-up only when the answer warrants it, branch around the section a respondent cannot judge, and keep the instrument's full coverage for the analysis side where it is needed. You keep methodological completeness while the person sees a shorter path.
Phone Layout Is Not an Afterthought
Assessments handed out in workshops get answered on the way out, in a car park, on a five-inch screen. The relevant test is not "does it fit" but "does the scale stay readable and does the submit control stay reachable". FormLM's fill layout aligns to the viewport rather than assuming width, and detects narrow portrait screens so the reading order stays single-column.
Measure the Right Loss
Before adding questions to raise rigour, look at the drop you would rather avoid: started-but-never-submitted. That number is the one that tells you whether the instrument is too long for the moment in which people meet it.