Everyone said they loved it. Nobody bought it.

You asked around. You posted it. You showed twenty people, and the replies were warm: this is great, I would definitely use this, let me know when it launches. Then it launched, and almost none of them did.

The natural reading is that people lied to you. Almost nobody did. The gap between what you heard and what happened is not dishonesty, and understanding what it actually is changes what you do next.

Praise is a sample, not a verdict

Start with the thing that is easiest to miss, because it is about who answered rather than what they said.

Everyone whose opinion you hold is somebody who chose to give it. They replied to your message, answered your post, stayed in the conversation. That is a group with a specific character: they were interested enough to engage, and in most cases they knew you, or knew somebody who knew you, or were the kind of person who answers a founder asking for help.

The people who considered your product and quietly decided against it did not reply. They had no reason to. Declining to buy something is not an event that generates a message, and explaining why you declined to a person who built it is socially expensive and returns you nothing.

So your feedback folder is not a reading of the market. It is a reading of the people who were willing to be in a conversation with you, which is a genuinely different population, and it is systematically warmer than the one holding your revenue.

They were honest. They were also guessing.

Now the harder half, which applies even to the replies you got from strangers with no reason to be kind.

When you ask somebody whether they would use a thing, you are asking them to predict their own future behaviour. People are reliably bad at this, not because they are careless but because the prediction is genuinely hard. At the moment of answering, they are imagining a version of themselves who has spare attention, an unambiguous need, no competing priority and no existing habit. That person does not exist on the Tuesday when your product actually needs choosing.

There is also a social cost to saying no to somebody's face. Not a large one, but large enough. “That sounds useful” ends a conversation pleasantly. “I would not use that” requires explaining yourself to somebody who has clearly put work in. Most people take the cheaper option, and they are not wrong to.

Some praise is real, and it matters. If somebody volunteered detail you did not prompt for, described a specific moment the thing would have helped, asked when they could pay, or came back unprompted a week later, that is a different kind of signal and it should not be discounted. The problem is not that positive feedback is worthless. It is that generic positive feedback and genuine intent look identical in an inbox.

The half you never collected

Put the two together and the picture is uncomfortable but clear. You have a warm sample answering a question they cannot answer accurately, and the group whose behaviour actually determines your revenue is not represented in your data at all.

This is why more feedback of the same kind does not help. Asking forty people instead of twenty gets you a larger version of the same sample, answering the same unanswerable question. The volume goes up and the information does not.

It is also why a founder can be genuinely diligent and still end up here. Nobody skipped a step. The step itself was measuring the wrong thing, and it felt exactly like doing the work.

Questions that return facts instead of kindness

The fix is not to stop talking to people. It is to stop asking them to forecast, and start asking them to remember.

What are you using for this now? This asks about a decision already made, with money, time and habit already behind it. Nobody has to guess and nobody can flatter you, because the answer happened before you turned up. If the answer is a spreadsheet, or nothing at all, you have learned something specific about what you are actually competing with.

When did you last have this problem, and what did you do? A real instance with a date attached is evidence. “This happens all the time” without one usually is not.

What did you try before that, and why did you stop? Abandoned solutions are the most honest thing in the conversation. People remember what annoyed them enough to quit.

Have you paid for anything in this area? Past spending is the strongest available proxy for future spending, and it is a fact rather than an intention.

We went deeper on this in the piece on what to ask and what to stop asking. The short version is that every useful question points backwards.

Where the missing half actually lives

The people who decided against you are not reachable by any survey you can send, because they are not on your list. But they are not silent. They just left their evidence somewhere other than your inbox.

They searched something before they found anything, and what people search is public. They read reviews of the alternatives and sometimes wrote them, and reviews of a competitor are more candid about your category than anything ever written about you. They asked in communities where your buyers already are. They bought something else, and what sits beside you on a shelf is visible.

None of that is written with you in mind, which is exactly why it holds. It is also scattered, unweighted and slow to read by hand, which is the only real reason most founders never read it.

That is the job MarkLens does: read those outside sources together, grade how much evidence stands behind each finding, and say plainly where the signal ran out. It will not tell you that everyone loves your product. Your inbox already did that, and it was the least reliable thing you had.