The First Ninety Days

Onboarding programmes end at week one or week two. The departures do not.

The second peak

A first spike in the first fortnight, which everybody knows about.

And a second, smaller and more spread, between the first and third month.

Which is a different group leaving for different reasons: the novelty has gone, the ramp pressure is real, the first payslip has landed, and the person now knows what the job is rather than what they hoped.

Almost no operation has anything scheduled in that window.

What is happening at each point

Week four: the person is close to standard and frustrated at not being there. The ramp curve conversation matters most here.

Week six: they have worked a full cycle of the rota, including the parts they had not experienced — the weekend, the night, the peak Monday.

Week eight: they have seen how absence, holiday and overtime actually work, as opposed to what the handbook said.

Week twelve: many agency contracts and probation periods land here, and the person is deciding whether to commit.

The scheduled contacts

Week four: the ramp conversation, with the curve and the number.

Week eight: a longer conversation about how it is going, away from the floor.

Week twelve: a formal review with a decision and a statement of what comes next.

Three contacts, each twenty minutes, and they cover the window where nothing else does.

Scheduled in the system rather than left to a supervisor's memory, because supervisors under pressure will forget and the person will notice.

What the week-eight conversation should cover

Is the job what you expected by now?

Is the shift working, including the parts you had not done when you started?

Is anything hurting?

Do you know what you could move toward?

And is there anything that would make you look elsewhere?

The last question again, plainly asked, produces answers that are almost always specific and frequently fixable.

Probation, handled properly

A period that exists in the contract and is mostly ignored until somebody wants to use it.

Which is unfair to the person and useless to the operation.

Use it as a two-way review: a stated decision, on stated criteria, with the criteria known from day one.

And where somebody is not going to make it, say so early rather than extending by silence, which produces a bad leaver and a grievance risk.

The holiday and pay milestones

The first holiday request, which tests whether the process works and is a common source of frustration.

The first full pay period, which is when people finally see what they actually earn.

And any progression point in the pay structure, which should arrive before the twelve-week decision rather than after it.

Time the pay structure to the attrition curve, which is a design decision nobody makes deliberately.

Handing over

At ninety days the person stops being a new starter and becomes an employee, and the scheduled attention stops.

Which is correct, and it should be explicit: a conversation marking it, and a statement of what the normal rhythm is from here.

People remember being told they had passed something, and almost nobody is told.

The measure

Attrition by week across the first twelve, as a distribution.

Scheduled contacts held against contacts due, which is the check on whether the programme is real.

Probation decisions made on time.

And ninety-day survival, which is the headline number this whole collection is built around.

Putting it in the system

Three calendar entries per starter, created at offer.

Assigned to the supervisor, visible to whoever owns retention.

Counted weekly against due.

Left to memory, they happen for two months and then stop, which is the drift this whole collection keeps returning to.

Timing the pay structure

Any progression point should arrive before the twelve-week decision rather than after it.

Which is a design choice nobody makes deliberately, and it is free to change.