FOUNDER FIELD GUIDES
How to get useful feedback on a SaaS product without an audience
Find someone who has done the task, ask about a real experience, and use their answer to decide what to test next.
By the Founder Battle team · Published · Updated · 6 min read
A friendly “looks great” is encouraging. It rarely tells you what to fix. Useful feedback identifies a specific person, a task they tried, where they hesitated, and what happened next. You can collect that evidence without a launch-day audience.
Start with a decision, not a request for opinions
Write down the decision you need to make: “Can a new visitor understand the offer?”, “Can an accountant finish the first import?”, or “Which of these two products would a potential buyer try first?” Each needs a different kind of evidence. If you ask everyone “What do you think of my SaaS?”, you will receive answers that are hard to act on.
- Name one target user and the job they are trying to do.
- Choose one observable task or comparison.
- Write down what would change your next product decision.
- Decide how many conversations you can actually follow up on.
Find people who have the problem
Start with current trial users, people who abandoned onboarding, or prospects who recently tried another solution. A small relevant group is more informative than a large group of other founders who would never buy your product. If you have no users, recruit in a community where the problem is discussed and ask permission to learn about the workflow before sharing your link. For help choosing a channel and screening volunteers, use our guide to finding beta testers for your SaaS.
Screen for experience rather than enthusiasm. “When did you last do this task, and what did you use?” is more useful than “Would you use my app?” This follows Nielsen Norman Group’s guidance on recruiting relevant participants and asking about past behaviour.
“Hi [name], you mentioned [task] in [context]. I’m building a tool for people who do that. Would you have 15 minutes to walk me through the last time you did it? A sample file is fine. I’d like to understand where the work slows down.”
Use three signals for three different questions
| Signal | Good for | It does not prove |
|---|---|---|
| Watch someone attempt a task | Finding confusing steps, missing information, and moments of hesitation. | How common the problem is across your whole market. |
| Short user interview | Understanding the job, alternatives used, and buying constraints. | What someone will actually do next month. |
| Head-to-head product verdict | Learning which of two public product pitches a participating voter prefers, with reasons where provided. | Usability, retention, revenue, or a statistically representative market preference. |
Interviews report what people remember and say; observing use reveals behaviour. Combining them helps you distinguish a persuasive pitch from a product that works in the buyer’s workflow. NN/g explains the strengths and limits of interviews.
Product feedback questions for a first user conversation
Choose a question that could change a decision you face this week. If a new user stops during setup, ask about the step where they stopped. Save pricing or feature questions for someone who has tried the product and can describe what they used instead.
| What you need to learn | Ask the user | Useful follow-up |
|---|---|---|
| What prompted the search | “What happened the last time you needed a tool for this?” | “What did you try first?” Listen for a specific event you could recognise in another customer. |
| Whether the pitch is clear | “From this page, what would you expect to be able to do?” | “Which part gave you that impression?” Their answer points to the words or image that set the expectation. |
| Where setup breaks down | “What did you expect to happen when you clicked there?” | “What happened instead?” Watch the attempt before explaining the screen. |
| What they use instead | “How did you finish the job last time?” | “Could you show me the steps using an example?” A spreadsheet or a colleague may be the real alternative. |
| Whether the product helped | “What did you do with the result after you left the app?” | Ask for the next step. Sharing a report, redoing it by hand and leaving it unopened suggest different changes. |
| Why they stopped using it | “The next time you needed to do this, what did you use?” | “How did you make that choice?” Leave room for the possibility that the task never came up again. |
For a short in-app survey, ask one of these immediately after the relevant action. After a report export, “What will you use this file for?” is easier to answer than a request to rate the whole product. An interview gives you room to ask what happened the last time that person exported a file.
How one answer can change what you build
Imagine a freelancer trying a proposal tool with a sample client. They finish the proposal, click “Share”, then stop. They ask for a PDF button. Before adding an export feature, ask: “How did you send your last proposal?” Suppose they say, “I emailed a PDF because I knew my client could open it.” Now there are two plausible needs: a file for their records, or confidence that the recipient can open the link.
Give them a sample recipient link and let them open it in a signed-out browser. If they send the proposal after seeing what the client sees, try making the recipient preview easier to find and watch the next freelancer use it. If they still need a file for procurement, ask to see a blank example of the document the buyer requires. The same request for a PDF leads to two different product decisions.
How a Founder Battle verdict fits
A Founder Battle challenge puts two public products side by side. Arena Votes from people who discover the battle inside the Arena determine a valid result. Votes arriving through a founder’s shared link are displayed separately as Support Votes; they do not decide the winner or move the rank. Structured verdict reasons can suggest what to investigate next.
For example, in the public Lyntiq AI vs ProposalForge AI battle, the recorded Arena score was 5–2 for Lyntiq AI, while ProposalForge AI had three Support Votes. The result followed the Arena score. That demonstrates how the two vote sources are separated; seven Arena choices are not proof that one product is better for every buyer.
Judge-pool composition, small samples, presentation quality, and the people who choose to participate can still affect a verdict. Treat a battle as a prompt for follow-up: Which promise was clearer? Which task should you test with your target users? What needs changing before a rematch?
Turn feedback into one decision
After each session, note the person’s role, the task, what you observed, their own words, and one possible change. Group repeated problems by task rather than counting positive comments. Choose one change you can verify with another user. If feedback conflicts, check whether the people had different jobs or levels of experience.
Target user: ____ · Task attempted: ____ · Observed obstacle: ____ · What they said: ____ · Proposed change: ____ · How we will check it: ____
When you have a public product ready for a direct comparison, submit it to Founder Battle. Submitting and challenging are free. A paid boost or sponsorship can buy labelled exposure, never votes, an outcome, or rank.