How to Run a Survey to Validate Your Startup Idea

What a survey can and can't tell you

A survey is good at one thing: measuring how common a problem or opinion is across a lot of people, quickly and cheaply. That is its strength and its trap. Because it scales, it feels like hard data, but a survey only captures what people are willing to type into a box, which is often not what they will actually do or pay for.

Use a survey to spot patterns: which problem comes up most, what people currently use, what they spend, where they get stuck. Do not use it to ask 'Would you buy this?' and treat a wall of yes answers as a green light. People are polite, optimistic, and bad at predicting their own future behaviour. The most useful survey questions are about the past and the present, not the future.

Think of the survey as one instrument in a kit. It works best after a few real conversations have given you a hunch, and before you build anything. If you are still figuring out the difference between testing the problem and testing your solution, the piece on problem validation vs solution validation is worth reading first.

Get clear on what you're trying to learn

Before you write a single question, write down the decision the survey will help you make. 'I want to validate my idea' is too vague. A sharper version: 'I want to know whether small clinics in my city track appointments manually, how much time it costs them, and what they currently use.' Now every question either serves that decision or gets cut.

Surveys go wrong when founders treat them as a fishing trip and cram in twenty questions while they have someone's attention. You end up with a long form nobody finishes and a pile of data you can't act on. Pick two or three things you genuinely don't know and that would change your plan. If an answer wouldn't change anything you do next, the question is decoration.

• The problem: do they actually have it, and how often?
• The current solution: what do they use today, including spreadsheets, WhatsApp, or nothing?
• The cost: time, money, or frustration tied to the problem.
• The customer: who they are, so you can tell if the right people answered.

Write questions about behaviour, not opinions

This is the single biggest factor in whether your survey is useful or misleading. Opinion questions ('Do you think an app like this would be helpful?') invite flattering, meaningless answers. Behaviour questions ('When did you last face this problem, and what did you do about it?') force people to recall reality.

Anchor questions to the last time something actually happened. Ask what they did, what it cost them, and whether they tried to fix it. If someone has cobbled together a workaround, paid for a clumsy alternative, or built a messy spreadsheet, that is a strong signal the problem is real. If they have never done anything about it after facing it many times, the pain probably isn't sharp enough to build a business on.

Lead with multiple-choice and rating questions because they are fast to answer and easy to count. Add one or two open text boxes at the end for the gold: people describing the problem in their own words, which becomes your marketing copy later. The same instinct applies in live interviews, covered in how to do customer interviews.

• Good: 'In the last month, how many times did you do X manually?'
• Good: 'What do you currently use to handle this?' (with real options plus 'Other')
• Good: 'Roughly how much do you spend on this per month?' (offer ranges)
• Avoid: 'Would you use a tool that does X?'
• Avoid: 'How likely are you to pay for this?' before you've shown them anything real

Keep it short and write neutral questions

Long surveys get abandoned. Aim for something a busy person can finish in two to three minutes, which usually means around five to ten focused questions. Every extra question lowers your completion rate and the quality of what you get, because tired respondents start clicking randomly.

Watch out for leading questions that fish for the answer you want. 'How much do you struggle with messy invoicing?' assumes they struggle at all. 'How do you currently handle invoicing?' lets them tell you the truth, including 'it's fine.' You want the truth, even when it's disappointing.

Order matters. Open with one easy, relevant question to pull people in. Put screening questions near the start if you need to filter, and save the open-ended question for last, when the respondent is warmed up. A free tool like Google Forms is all you need; the platform barely matters next to the quality of your questions.

1. Open with a one-question screener to confirm the respondent is your target customer.
2. Ask how they handle the problem today (behaviour, not opinion).
3. Ask what is annoying or costly about their current approach.
4. Ask how often the problem comes up and roughly what it costs them.
5. Leave one open text box for anything else they want to add.
6. Optionally collect an email only from people who want to hear more, which becomes a warm list.

Reach the right people, not just any people

A survey full of responses from the wrong audience is worse than no survey, because it gives you false confidence. Fifty answers from people who match your real customer beat five hundred from random friends and family.

Go where your target audience already gathers. That might be a specific subreddit, a niche WhatsApp or Telegram group, a LinkedIn audience if you sell to professionals, an industry forum, a college, or a local market association. For offline businesses, sometimes the fastest path is standing in the right place and asking people directly. Always include a screening question so you can separate real prospects from everyone else during analysis. This pairs well with broader market research on a budget, which leans on the same instinct: go to the source.

Be honest about friends and family. They want to encourage you, so their answers skew positive. Their data can still help if they genuinely fit your customer profile, but tag them so you can look at their responses separately. If you are running this on a shoestring, the approaches in how to validate a startup idea without spending money pair well with a free survey.

How many responses do you actually need

There is no magic number, and you do not need statistical significance to make an early decision. For most pre-launch validation, somewhere between thirty and a couple of hundred responses from the right people is enough to see clear patterns. If a problem is real, it tends to show up loudly and early.

Watch for the point where new responses stop telling you anything new. When the tenth person describes the same workaround the first nine did, you have your signal. That repetition is more convincing than a large sample of vague agreement.

Treat survey numbers as directional, not precise. A survey can tell you 'most people in this group still do this by hand' with reasonable confidence. It cannot tell you 'exactly 37 percent will convert.' Don't dress up rough early data as something more exact than it is.

Read the results without fooling yourself

The danger after collecting responses is confirmation bias. You want the idea to work, so you notice the encouraging answers and skim past the warning signs. Read the open-text answers carefully, especially the negative and indifferent ones. A shrug is data.

Look for three things. First, intensity: is the problem a daily irritation or a once-a-year annoyance? Second, current spending: people already paying time or money to solve it are far more promising than people who say it bothers them but have done nothing. Third, language: the exact words people use to describe their pain are a gift, both for refining the product and for writing copy that resonates.

If the results are lukewarm, that is a win, not a failure. You just saved months of building something nobody needs. Adjust the idea, the audience, or the problem and run a tighter survey again.

What to do after the survey

A survey narrows your focus; it doesn't close the loop. The strongest signal is someone giving you something scarce: their time, their email, or their money. So follow the survey with a sharper test. Start with real conversations, because a ten-minute chat surfaces nuance no form can.

Then test the part a survey can't: whether people will act. Build a simple page describing the solution and see if respondents sign up, an approach covered in how to build a landing page to test demand. Better still, try to pre-sell or get a soft commitment before you build anything. Real demand shows up when people act, not when they tick a box.

When the signal is consistent across surveys, conversations, and a real ask, the goal is to ship the smallest thing that tests your core assumption, not a feature-packed product. That is exactly the kind of focused first build Xolver exists to help with.