The honest feedback loop — how good vibe coders actually iterate

Ask your friends what they think of your app and they'll tell you it's great. They're not lying, exactly. They want you to feel good, and they have no idea how to be useful, so they default to encouragement. That encouragement feels lovely and teaches you nothing, and a lot of apps die having only ever heard it.

Getting better needs the opposite: feedback from people who have no reason to be nice to you. That's uncomfortable, which is exactly why it works. Let me walk through how to get it, and how to tell the useful kind from the polite kind.

Why the nice feedback is the useless feedback

The closer someone is to you, the more their feedback is about protecting your feelings and the less it's about your product. Your mum thinks it's wonderful. Your friend says "yeah, I'd use that" and never opens it again. None of this is malicious, and all of it is noise.

The reason it's dangerous isn't that it's wrong. It's that it feels like validation. You walk away encouraged, having learned nothing, and you keep building in a direction nobody actually wants. Polite feedback is more harmful than harsh feedback, because harsh feedback at least points somewhere.

Go to strangers in your audience

Useful feedback comes from people who match your target user and owe you nothing. They'll tell you it's confusing, because they're not trying to spare you. They'll bounce off the thing that you thought was obvious. That sting is the signal.

You find these people in the same places you find users, the niche communities where your audience already gathers. Post the thing, ask for honest reactions, and brace yourself. The builder who can sit with "I didn't get what this was for" and treat it as gold is the builder who improves. The one who explains why the stranger is wrong learns nothing.

Watch what they do, not just what they say

Here's the most reliable trick: people's actions are far more honest than their words. Someone might say "nice app," then never sign up. Someone might say nothing and use it every day. The words are polite; the behaviour is true.

So whenever you can, watch instead of asking. Where do they hesitate? What do they click first? Where do they give up? You don't need fancy tooling to start, even sitting beside one person while they try it, saying nothing, tells you more than ten "what do you think" surveys. When you do ask, ask specific questions, "what did you expect to happen when you clicked that," not "do you like it."

Make it a loop, not a verdict

The mistake is treating feedback as a single exam you eventually sit. Build it into a rhythm instead: ship a small version, watch how a few real people use it, ask one or two sharp questions, make one change, ship again. Small and frequent beats big and rare, because you can actually act on small.

Each turn of that loop costs little and teaches a lot. Over a few weeks it pulls your product toward what people actually want, which is a very different place from where your friends said it was already great.

The honest-feedback half is one of the two things Shipyard was built for: an invite-only place where real builders review your work and tell you the truth, not the polite version. We're opening the first 50 founding spots now. If the feedback you're getting is all suspiciously kind, come and get the honest kind.