A pattern I stopped being able to unsee
Over the years I have been handed a fair number of stalled projects, the kind where a team hands you something half-built and slightly on fire and asks you to get it moving again. They came from different industries, different stacks, different sizes. And underneath, they had failed in almost exactly the same way.
It was never the thing everyone pointed at. Not the tech choice, not the difficult client, not the missed deadline. Those were symptoms. The root cause, every single time, was that nobody could tell me what had actually been decided, and why.
I resisted this conclusion for a long time, because it is almost too simple. Surely the real problems were the hard technical ones, the difficult stakeholders, the scope that kept moving. But every time I dug, those turned out to be things the team could have handled fine if they had shared a clear picture of what was true. The missing record was not one problem among many. It was the thing that turned every other problem into a crisis.
The shape of the failure
You arrive, you ask the obvious questions. Why is it built this way. Who agreed to this scope. What did the client actually sign off on. And you get shrugs, or three different answers, or a confident answer that turns out to be one person's memory. Decisions had been made, plenty of them, but they lived in people's heads, in scattered chat threads, in a call nobody wrote up. There was no shared record of what was true.
Once that is gone, everything downstream rots quietly. People rebuild things that were already decided. Two parts of the team proceed on contradictory assumptions and nobody notices until they collide. The client remembers agreeing to something different from what got built, and there is no record to settle it. Ownership evaporates, because if no one wrote down who owned a decision, no one owns its consequences either.
What makes it so corrosive is that none of it announces itself. There is no alert for a decision that lives only in one person's head. The project keeps moving, feeling productive, while underneath it the shared understanding quietly dissolves, until one day two things built on different assumptions meet, and suddenly nobody can agree on how you even got here.
By then it is not really a documentation problem anymore. It is a trust problem. People start defending their own version of events, because their version is the only record they have, and the project turns from building a thing together into arguing about what the thing was ever supposed to be. You cannot ship your way out of that. You have to rebuild the shared truth first, which is slow and thankless and the only thing that works.
A project doesn't stall because a decision was wrong. It stalls because nobody can remember which decision was made.
Why it is so common, and so fixable
It happens because keeping the record feels like overhead when things are moving fast, and moving fast is when it matters most. It is the first thing dropped and the last thing missed, right up until the project seizes and you cannot move because you no longer know where you stand.
The fix is unglamorous and nearly free: a single, current, boring place where decisions live, with who made them and why. Not a heavier process. Just one version of the truth that outlives any one person's memory. Almost every rescue I have done started there, with rebuilding the record before touching a line of code, because you cannot un-stall a project whose own history is in dispute.
The question worth asking now
The reason I keep writing this down is that the fix is so cheap and the failure is so expensive, and almost nobody does it until after they have been burned. It never feels urgent. It is the sort of thing you promise to sort out once the current crunch is over, and the crunch is never over, and then one day it is the reason everything stopped.
The teams that never stall are not the ones with the best tools or the smartest people. They are, boringly, the ones who keep the record: who decided this, when, and why. It is the least glamorous discipline in delivery and the one that most reliably separates a project that can absorb a shock from one that quietly comes apart at the first real disagreement.
So if you are on something that feels stuck right now, the question worth asking is not what is wrong with it. It is this: could any of you, right now, agree on what was decided and why, and if not, is that not the actual problem?