Why Clients Change Their Minds Right Before Launch

We spent about a month building a product for a client. They were with us the whole way, testing as we built, sending feedback, signing off on each piece as it landed. Everything you could point to said we were aligned.Then, a few days before launch, they told us this was not the product they had in mind. Not a tweak. A different flow, a different shape, a whole new idea of how the thing should work, and they laid it out in full. The project reset into a near-complete redesign, right at the finish line.My first reaction was the one every developer has had at some point: you had a month, why now. But that reaction is wrong, and it is an expensive thing to get wrong. The client was not messing us about. They were doing the one thing any of us can do, which is know something is off the moment they finally see it in front of them.Nobody knows what they want in the abstractPeople cannot judge a description. They can only judge an experience. We keep forgetting that.When a client signs off on a wireframe or a spec, what they are approving is an idea of the product. The real thing does not exist yet, so they picture it, and the picture is always generous. It fills in the gaps and quietly assumes the whole thing will feel right. It is only when the product is real enough to actually use that they can see the distance between what they imagined and what they got, and that distance tends to be huge, because the version in their head was never as solid as their sign-off made it look.So “the client changed their mind at the end” is usually nothing of the sort. It is their first honest reaction, turning up at the first moment an honest reaction was even possible. They did not know what they wanted. They could not have. For most people, wanting is something that happens in response to a thing in front of them, not a plan they carry around in advance.Our real mistakeAnd once you look at it like that, the failure was ours, not theirs.We only put something concrete enough to really react to in front of them at the very end. They spent a month testing pieces, but a piece does not set off the reaction that the whole product does. It was only when it all came together, days before launch, that they could feel the shape of it, and the shape was wrong. The feedback did not arrive late because they were careless. It arrived late because we did not hand them the thing that produces that feedback until it was nearly too late to do anything with it.We had built the whole process around making the product. We should have built it around getting that real reaction out of them as early and as cheaply as we could.The excuse just disappearedFor most of my career there was a genuine defence for working this way. Getting a usable version in front of a client early was expensive. Throwaway prototypes felt like waste, so you built the real thing and hoped the sign-offs would hold.That defence has gone. AI has made it cheap to throw together a rough, clickable, deliberately disposable version in days instead of weeks. Something the client can move through and feel while it still costs a conversation to change direction rather than a redesign. The entire point of it is to draw out the “no, not that” while it is still cheap to hear.Which means finding out about the wrong flow days before launch is not something the tools force on you anymore. It is a choice now, made by teams who are still saving the first real experience for the end.What I do nowI get something usable into their hands as early as I can, and I make a lot of noise about it being disposable, so nobody mistakes a rough prop for a commitment. I treat any sign-off on a document as provisional, because it is agreement with an idea, not with the product. And I assume the real requirements will not show up until the client has touched something real, so I build the timeline around reaching that moment early instead of bracing for it late.Clients not knowing what they want is not the problem. It is the ordinary, permanent condition of building things for other people. The only real question is whether you let them find out on day three or day thirty.