Reading the Tenure Curve
Time and activity data need context: workload, training, equipment and shift design all change the number. If the relevant page is considered, publish the purpose, minimise collection and review whether the information actually improves the process.
A useful independent reference for this topic is CIPD. Treat this article as general operational material and confirm any legal or safety decision with current official guidance.
One chart, drawn from payroll records in an hour, that locates the problem more precisely than any survey or benchmark.
An annual attrition percentage describes a mixture. The distribution of length of service at leaving describes a cause.
Drawing it
Take every leaver over twelve months.
For each, calculate weeks between start date and leave date.
Plot the count by week, from one to fifty-two, then monthly beyond.
That is the whole method. Payroll holds both dates and the chart takes an hour.
Reading the shape
A spike in week one or two: induction, first-day logistics, equipment, or an advert that described a different job.
A spike at weeks three to five: the ramp, the first payday, a training rate that was not disclosed, or the physical demand becoming real.
A rise around week twelve: probation decisions, agency contract points, or the moment people decide to commit and do not.
A steady rate across the first year: pay against the local market, or a general condition rather than a moment.
A rise at twelve to eighteen months: absence of progression, which is the second-order problem to fix once the first is under control.
The cuts that matter
Direct against agency, always, because they have different curves and pooling them hides both.
Peak against permanent.
By shift pattern, which usually shows one pattern losing people faster.
By team, which is the supervisor question.
By source, which is the recruitment question.
Each of these is one additional field and each splits a single ambiguous chart into two informative ones.
What the curve does not tell you
Why, which comes from the exit conversations and the stay interviews.
The curve locates the window; the conversations explain it.
Using one without the other produces either a precise diagnosis of nothing or a rich explanation of something that may not be the main problem.
Comparing across time
Draw it again after any intervention, over the same period length.
The question is whether the spike moved or shrank, not whether the annual percentage changed — which is a lagging figure contaminated by everything else.
A first-fortnight spike halving is visible within three months and is the fastest feedback available in this subject.
Benchmarks
Industry attrition figures are of limited use because they mix operations with completely different mixes of agency labour, peak volumes and shift patterns.
Your own curve last year is the only comparison worth making.
And an internal comparison between your sites, provided their agency proportion and peak profile are similar enough to be comparable.
The version to show people
A simple bar chart of leavers by week of tenure.
It is immediately legible to people who will not read a table, and it makes the argument for spending on the first fortnight without anybody having to make it verbally.
Put it in front of whoever controls the induction budget.
The measure
The curve itself, redrawn quarterly.
The proportion of leavers who went within thirty days, which is the single headline number it produces.
Median tenure at leaving, by cohort.
And the change in the shape after an intervention, which is what tells you whether anything worked.
Drawing it the first time
Export start and leave dates for twelve months of leavers.
Calculate weeks, count by week, chart it.
An hour, and it will change where the next conversation about retention starts.
What to do with the shape
A week-one spike: the first day, the equipment, the advert.
Weeks three to five: the ramp, the first payday, the training rate.
Week twelve: probation and agency contract points.
Flat: the rate against the local market.
Four shapes, four different programmes, and drawing the chart is what tells you which one you are running.
How it fails in practice
The annual percentage is reported, it mixes first-week leavers with five-year leavers, and every discussion about it becomes a discussion about pay..
The check
Draw the curve once.
It takes an hour from payroll records and it changes where the next conversation starts.