Standard startup advice always hits the same note: talk to your customers. But while face-to-face interviews feel productive, they are often a trap. Users are usually too polite to tell you your idea is bad. They struggle to explain what they actually need, and they almost always agree to hypothetical prices for products they will never actually buy.
Real product discovery means moving away from asking for opinions and moving toward observing behavior. You need proof of frustration and a clear intent to pay, not just a verbal pat on the back. Here is how founders are validating ideas without wasting weeks on Zoom calls.
Review Mining and the Power of Unbundling
The most honest feedback doesn't happen in a scheduled interview. It happens when a user is frustrated enough to vent online.
Most enterprise software is a bloated mess. Companies pay thousands of dollars for a suite of 50 tools when their employees only use two. This creates a massive opening for "unbundling"—taking one popular feature out of a complex, expensive suite and building a polished, standalone version of it.
To find these gaps, founders shouldn't look at five-star praise or one-star rants. The real insights live in the three-star reviews on sites like G2, Capterra, or the App Store. These reviews usually sound like this: "I love the reporting tool, but the rest of the software is too slow and expensive."
That complaint is a ready-made product roadmap. If you find dozens of people saying they only keep a Salesforce subscription for one specific tracking feature, you have a validated concept for a micro-SaaS.
Social Listening on Reddit and X
People talk differently on social media than they do in surveys. On Reddit or X, the language is raw and the problems feel more urgent.
Instead of asking broad questions to a general audience, use specific search operators in niche subreddits. Look for phrases like:
* "How do I..."
* "Alternative to [Competitor]..."
* "Why is [Process] so hard?"
The real "gold" is a thread from three years ago that people are still commenting on today. If a problem has remained unsolved for that long, it’s a persistent pain point that people are desperate to fix.
Smoke Tests: The "Fake Door" Method
Interviews might help you understand a problem, but they rarely prove someone will pay to solve it. A smoke test fixes this by testing purchase intent before you write any code.
Build a simple landing page for your potential product. Clearly describe what it does and what it costs. When a user clicks "Sign Up" or "Buy Now," send them to a page explaining that the product is in beta or launching soon, and ask for their email address.
The conversion rate on that "Buy Now" button is the only data point that matters. It separates the people who say they like your idea from the people who are actually willing to reach for their wallets.
Automating the Search for Friction
Manually digging through thousands of G2 reviews is effective, but it takes weeks. New tools are emerging to speed up this workflow.
For example, Feature2Product uses AI to scan thousands of software reviews across different markets. It specifically hunts for those unbundling opportunities—the high-demand features buried inside clunky enterprise suites.
Rather than guessing which features are worth building, the tool provides a "productizability" score and estimates the effort required to build it. It turns months of manual validation into a data-backed starting line, helping indie hackers move straight to the build phase with confidence.