On this page
Consider what one shift learns. The ice machine struggled around four. A supplier delivered short and the replacement is due Thursday. A large group behaved badly and may book again. One of the new starters is picking things up faster than expected. A regular asked about something not on the menu, for the third time.
Now consider what reaches the next shift. Usually the ice machine, because it is still broken. Sometimes the delivery, if it affects tonight. Rarely the rest.
None of that is negligence. The handover happens at the busiest possible moment, between people who are tired and people who are about to be, and the material that gets passed is whatever is urgent rather than whatever is important.
Why this is the expensive kind of loss
Losing a fact is cheap. Losing a pattern is not, and patterns are made of facts nobody recorded.
A supplier that short delivers once is an incident. A supplier that short delivers on eleven occasions across a year is a commercial problem worth raising, and no bar that fails to record the incidents will ever see the pattern. It will simply experience a low grade irritation for years and never quite decide to act on it, because nobody can say how often it happens.
The same is true of the guest who asks for something not on the menu, the section that runs slow on the same night each week, and the product that goes out of stock more than it should.
The three places it leaks
The handover
The largest leak, for the reasons above. It happens daily, so a small percentage loss compounds quickly.
The undocumented decision
Something is noticed, a change is made, and the reasoning is not recorded. Six months later the change is reversed by someone who does not know why it was made, the original problem returns, and the business has paid for the same lesson twice.
The departure
The visible one, and the one that gets attention precisely because it is a discrete event with a date attached. It is usually the smallest of the three, because it is the accumulated total of the other two finally becoming legible.
What a fix looks like
The instinct is to require more documentation. That fails, reliably, because it adds work to the shift that is already the most compressed part of the day, and it competes with actual service.
The version that works has to be a by-product of running the place rather than an additional task. Three properties matter:
It has to be captured where the work happens, not in a separate system somebody has to remember to open.
It has to be searchable later, because the value is in seeing the pattern across eleven months, not in reading yesterday's note.
It has to include what was decided and what happened, not just what occurred. An observation without an outcome is half the record, and the missing half is the part a successor needs.
Why this connects to retention
A venue that loses knowledge continuously is harder to work in. Every shift starts by reconstructing context. The same problems recur because nothing that was noticed reached anyone with the authority to change it. That is a tiring environment, and tired people leave, which removes more knowledge.
One venue in our case studies reported a 73 percent reduction in turnover attributed to LearnTAP, a client reported figure from a single site rather than a benchmark. Whatever the size of the effect elsewhere, the direction of the mechanism is worth taking seriously: knowledge retention and staff retention are not two problems.
If you want the underlying study rather than the argument, it is on the research page.