Every additional question is a chance for a respondent to leave. A few changes that reliably improve completion:
Open with short, low-effort questions (rating, single choice) before asking for open-ended text - respondents who've already answered a few questions are more likely to finish than ones facing a blank text box first.
Only mark a question Required if you genuinely can't use the data without it. Every required field is a point where a respondent can get stuck and abandon rather than skip past.
If a screening question makes the rest of the survey irrelevant to some respondents, use skip logic to end the survey there rather than making them click through questions that don't apply to them.
A survey description that says "5 questions, about 2 minutes" measurably reduces abandonment compared to an open-ended survey with no stated length - people are far more willing to finish something with a known, small cost.
Your completion rate tells you if there's a drop-off problem at all; check individual responses for where in your own survey people are stopping to find the specific question that's losing people.