The Two Weeks Before Someone Quits

Someone hands in their notice and the reaction is always the same: "I had no idea."

They usually decided weeks ago. And in most venues the decision left a trace in the rota before it reached anyone's inbox.

What the trace looks like

Not one signal. A shape, made of ordinary things that are unremarkable alone.

Shifts stop being picked up. The person who always took the extra Friday stops taking it. Nothing was refused — they simply stopped volunteering.

Swaps go one direction. They give shifts away and stop taking them back. The ratio is the signal, not the count.

Availability narrows. The window they say they can work quietly shrinks, often to exactly the shifts they cannot avoid.

Requests cluster on weekdays. Because the interviews are on weekdays.

None of these is a resignation. Together, over a fortnight, they are a fairly reliable one.

Why nobody sees it

Because each piece sits with a different person.

The swap went through a group chat. The availability change was a text to a manager. The unclaimed extra shift is an absence, and an absence is the hardest thing to notice — nothing happened, so nothing was recorded.

A resignation is invisible not because it is subtle but because it is distributed.

What is worth doing about it

Here is where most advice goes wrong. The instinct is to intervene and change their mind.

By the time the pattern is visible, that is usually too late — the decision is generally made before the behaviour changes, not after. Treating it as a save-the-employee moment mostly produces an awkward conversation and a counter-offer that fails within three months.

What the warning is actually worth is time.

Time to start recruiting before you are a person short on a Saturday.

Time to move the handover — what they know about specific guests, which regulars ask for them, what is half-finished. This is where the real loss is, and it can only be captured while they are still there.

Time to find out why, honestly, while there is still goodwill. Not to change their mind — to learn whether the reason will take the next person too.

The reason is usually structural

When people leave a venue, the stated reason is often money and the actual reason is often the rota.

Shifts that change late. Being scheduled on the nights they said they could not work. Being given the shifts nobody wants without it ever being acknowledged. These are the ones that accumulate quietly and produce the fortnight.

⚠️ That means the signal is worth reading in aggregate, not just per person. If three people in six months showed the same pattern, the answer is not three conversations — it is the schedule.

The count that matters more than the rate

Turnover percentage is a comfortable number because it goes into a report and nothing happens.

The useful figure is how many of your leavers were people you wanted to keep, and whether that number is going up.

A venue with high turnover among people who were never going to stay has a hiring problem. A venue losing three people a year who were good has a very different problem, and the percentage does not distinguish them.

The trap in watching for it

Two of them.

Reading it as disloyalty. People leave. Treating the pattern as betrayal turns a two-week head start into a fortnight of bad atmosphere, and it guarantees the next person hides it better.

Acting on one signal. A single dropped shift is a person who was tired. The value is in the shape over two weeks, and using less than that produces accusations about nothing.

Three to hold

Swap ratio per person, rolling. Given away versus taken on. The clearest single indicator.

Extra shifts claimed, versus their own baseline. Compares each person to themselves — a low-hours person who was always low-hours is not a signal.

Leavers you wanted to keep, rolling twelve months. The number that tells you whether any of this is getting better.

Where the record has to sit

Swaps in a group chat, availability in texts, and the rota in a spreadsheet cannot be compared. That is why the fortnight passes unnoticed and the notice always feels sudden.

tasteck keeps shift submissions, swaps, and availability on the same record as the schedule, so who has been giving shifts away and not taking them back is a list rather than a hunch.

The notice was not sudden. It was just the first part that was written down.