Aviation is the only industry that institutionalised keeping a record of its own failures. When an Australian scientist designed the device that made the final minutes of a crashed aircraft audible, back in the 1950s, the idea he was selling was not a technical one: he was saying that after an accident everyone tells their own version, and the only impartial witness will be the recording. Today, when a plane goes down, the first thing anyone looks for is not the wreckage but that box.
Companies have no such box. Some of them believe they do.
Organisations archive their reputation, not their memory. Which is why they can make the same mistake three times and believe each time that it is the first.
Why the record isn’t kept
Looking at what an organisation records is looking at what it rewards. Every company has an archive of annual presentations: growth charts, accounts won, launches. That same company usually holds not one file on the projects that sank. Post-mortems happen, of course, but the people called into the room are already in a defensive posture, and within a fortnight the resulting document has turned into the draft of a success story. Most of what gets written under “what we learned” is in fact evidence that nothing was: we were too ambitious, the timing was wrong, the market wasn’t ready. None of these is a decision. They are weather reports.
Chris Argyris named this behaviour in “Teaching Smart People How to Learn,” published in Harvard Business Review in 1991. He showed that the brightest professionals had never learned from failure precisely because they had rarely failed; when something went wrong, their reflex was not to examine but to defend. Argyris called these defensive routines, and separated single-loop learning from double-loop: the first corrects the error, the second interrogates the assumption that produced it. Almost every organisation is excellent at the first. What the second requires is not intelligence but a record.
Because to question an assumption later, that assumption has to be written down somewhere. If you did not write what you believed at the moment the decision was taken, six months on your memory will tell you a version that vindicates the outcome, and nobody will be there to object to it. Everyone is acting in good faith. Nobody is lying. There is simply no box.
Whose record is it
The strange part is that a record accumulates even where none is kept. Teams now work by writing everything down: message channels, decision threads, comment lines, briefs that were cancelled. Nobody treats that accumulation as an archive, nobody indexes it, nobody is assigned to own it. It just sits there. A company’s real memory is this scattered heap that no one is responsible for.
And the heap has a price. The raw record of how an organisation runs (the order in which decisions were taken, what was tried and abandoned, which move was made against which number) is a saleable asset today. While the company stands, only the company can use the value of that record; when it no longer stands, someone else does. In neither case does the record disappear. What disappears is the discipline of looking at it.
Don’t try to turn this into a software project or a data strategy; taken there, it dies. What it takes is far smaller: beside every serious decision, a paragraph on what you believed while taking it. Not the numbers: the belief. “We expect this campaign to work because we think this audience wants this thing.” Read that paragraph again six months later and what you hold is not an outcome but a calibration: the only document that shows where your own judgement drifted. A team that does this for a year can no longer mistake the same error for a new one.
There is a cost: a written assumption cannot be hidden once it proves wrong. That is exactly the point. Organisations that keep no record are not forgetful; they are protected. And the cost of protecting an organisation from its own past is always paid by its future.
Nobody is going to fit the box for you.