Dear Colleague,

what you asked probably has a different answer for each and every questionnaire (from different questionnaires, different number of questions can be removed, assuming, that you have redundancy).

A good way to check this for you own questionnaire is by applying Cronbach's Alpha, and perhaps start removing questions here and there (what you think is not needed). You can also form a hypothesis about this; what do you think which questions are redundant / not important?

Of course, there are multiple other metrics and statistical tests for proving reliability, you might want to do a focused research on those.

I hoped I helped, kind regards,

Simon