Most customer research forms are a list of things the founder wants to know. That is the wrong starting point, and it produces the two characteristic failures: people leave halfway, and the ones who finish give you answers you cannot act on.
The fix is a single rule.
Ask about the past, not the future
"Would you use a tool that did X?" is the most common question on these forms and it is worth nothing. It asks somebody to predict their own behaviour in a hypothetical, and people are bad at that and polite on top of it. You will get a yes. The yes means they do not want to be rude.
"What did you do the last time this came up?" asks about something that actually happened. There is a real answer and they know it.
Every question below is a variation on that rule.
The six questions
1. What did you do the last time this came up?
The best question on any research form. Open text, and worth its cost even though open text is more work to read. The answers give you the words your landing page should use, which are never the words you would have chosen.
2. What did that cost you?
Money, time, or annoyance — let them pick which. This separates a real problem from a mild one, and it is the question that most reliably identifies who your customer actually is.
3. What did you try before that, and why did you stop?
The one people skip. Somebody who has already tried three things has a problem worth money; somebody who has tried nothing may not. And the reasons they abandoned previous attempts are the reasons they will abandon yours.
4. What do you spend on this today?
A price question about the past. Bracket it — free only, under $10, $10–25, $25–50, $50+ — rather than using a text box. Brackets are comparable across respondents and you avoid cleaning "20", "$20", "20/mo", "depends", and the person who typed a million as a joke.
5. Who would decide?
Reveals whether you are selling to a person or a company. Changes your entire go-to-market and is routinely omitted.
6. Anything I did not ask about?
Always last, always optional. A surprising share of the most useful sentences you will get arrive here, because it is the only question that does not constrain the answer.
Ordering, and where people leave
Two rules, and the second is the one that gets broken.
Cheapest question first. Whatever costs least to answer opens the form. Momentum is real and the first question is where the largest drop happens.
Do not ask for the email until you have earned it. Putting it first reads as a sales gate. Putting it last, after somebody has written three sentences about their situation, gets a much higher rate — they have invested, and the ask now looks like a natural end rather than a toll.
The exception: if the email is the only thing you actually need, ask for it first and stop. A research form and a signup form are different documents and merging them produces something that does neither job.
Why one question per screen matters more here than elsewhere
A fourteen-field grid makes the reader price the whole interaction before answering anything, at the moment they have the least investment and the least idea why any of it is being asked.
But the effect that matters more for research is subtler: a question alone on a screen can be written properly. In a dense grid every label is compressed to two words, because fourteen sentences would look absurd. Given a screen, question three can carry a sentence explaining why you are asking — and "why are they asking this" is what actually causes abandonment on the awkward question, not the typing.
The honest counter-case: if your form is three fields, one screen is better. Splitting three fields across three screens adds transitions to save nobody anything.
Read the abandonment
DevForm stores a partial response from the first answer rather than on submit, and the response inbox reports complete and unfinished counts separately.
That turns an abandoned form from an absence into evidence, and it is the most useful thing a research form produces. A cluster of drop-offs at one question means one of three things, and you can usually tell which by rereading the question:
- It asks for something they do not have to hand.
- It asks for something they are not ready to give, and you have not said why you need it.
- It is ambiguous, and stopping is easier than guessing.
None of that is visible in a completion rate. Partials expire after thirty days — somebody who abandoned a form did not agree to your keeping half their answers indefinitely.
Read every answer yourself
All of them, before you summarise anything, before you count anything.
The finding in a research form is almost never in the aggregate. It is in three people using the same unusual phrase, or one person describing a workaround you had not imagined. Both disappear the moment you tabulate.
If you have more than a couple of hundred responses, read a random hundred properly rather than skimming all of them.
What this needs
Free gives you 1 form, 10 questions, 100 responses a month — no conditional logic, no embed, no CSV export. Enough for a first research form, which is usually the shape above.
Starter, $9/mo adds logic, embedding, and export: 5 forms, 30 questions, 2,500 responses.
Not built, on any plan: file uploads, notification emails, webhooks, integrations, payments. Nothing arrives in your inbox when somebody responds — you check, or you export.