Every feature request you receive is a short story with most of the pages missing. Someone wants a CSV export. Fine. For what? For whom? Instead of what they do today? You can chase those answers by email, one person at a time, or you can ask for them once, on the form, while the person still cares enough to type.
The trick is asking for the right three things and stopping there.
1. What are you trying to do?
Not "describe the feature". People are bad at designing features and very good at describing their own afternoon. Ask what they were trying to get done when they hit the wall. "I need to send our roadmap to finance every quarter" tells you far more than "add Excel export", and it often points at a smaller fix than the one they asked for.
Put this question first and make it the big text box. Everything else is supporting detail.
2. What do you do today instead?
This is the question that tells you how much it hurts. A person who copies forty rows into a spreadsheet by hand every Friday has a different problem from someone who thought export would be nice to have. The workaround also tells you what "done" looks like. If they're pasting into a spreadsheet, a spreadsheet is the bar.
3. Who are you?
You need just enough to weigh the request later: their plan, the size of their team, maybe their role. Keep it to one or two fields, and make them choices rather than free text so you can filter by them.
If people sign in to post, you may already know most of this. Don't ask twice.
The one to leave out: "How important is this to you?"
Almost everyone picks the top of the scale. Of course it's important. They took the time to write it down. The answer tells you how the person felt in the minute they were typing, which is not the thing you need when you sit down to decide.
Importance shows up in better places. It shows up in votes from other people, in how many customers asked in their own words before you merged them, in what the workaround costs them, and in who is asking. You'll get a truer signal from any of those than from a slider.
Keep it short enough to finish
Every field you add costs you some of the people who would have finished a shorter form. Three questions is a good number. Five is the most you should ask on a public board, and only if the extra two genuinely change what you'd do.
A few practical things help:
- Mark the one or two fields you can't act without as required, and leave the rest optional.
- Write the placeholder as an example answer, not as an instruction. "We send the roadmap to finance every quarter" teaches better than "Describe your use case".
- Show similar posts while people type, so they can vote instead of writing a duplicate.
Then read what comes in
A form is a guess about what you'll need. After a month, read twenty posts in a row and ask which field you actually used. Drop the one you never looked at. Add the question you kept asking in replies. The form should get a little better every month, the same way the product does.