Where to Start
Before adding a dashboard, establish the baseline and the action it will trigger. A workplace platform such as find out more is more useful for a narrow, openly stated workflow question than for broad monitoring without an operational response.
The principles in “Where to Start” can be cross-checked against resources from Glassdoor, particularly where employment practice or warehouse safety requirements may have changed.
A first month that produces a diagnosis and two fixes, using data you already hold and costing nothing but attention.
Most retention work starts with a programme. The order that works starts with four numbers and a conversation.
Week one: the diagnosis
Draw the tenure curve from twelve months of payroll records.
Split it: direct against agency, peak against permanent.
Count last month's starters still present.
And read the advert, then walk the floor, and decide whether it is true.
Four things, one week, and they locate the problem.
Week two: ask the people who left
Twenty leavers from the last quarter, contacted a week or more after they went.
Five questions, by phone, by somebody neutral.
Response will be partial and the answers will be specific, which is the opposite of the exit form.
Group them. Two or three causes will account for most of it.
Week three: fix the two cheapest things
Whatever the diagnosis pointed at, take the two that cost least and do them now.
Typically: boots in the right size on day one, and a day-three conversation.
Or: an honest advert, and a message the evening before the first shift.
Doing something visible in the first month is what makes the rest of the work possible, because it demonstrates that the exercise produces change rather than a report.
Week four: set the measure and name the owner
Ninety-day survival, reported monthly, split by source and by agency status.
One person accountable for it, with authority across recruitment, induction and supervision.
And recruitment's measure changed from cost per hire to cost per ninety-day survivor, which is one change to one report and realigns everything downstream.
What to avoid
Starting with a pay review, which is expensive, slow and frequently not the cause.
Starting with an engagement survey, which measures the survivors and takes a quarter to produce.
A retention bonus, which pays people who were staying anyway.
And a programme with a name, which invites scepticism before anything has been demonstrated.
The second month
The day-three and week-four conversations scheduled in the system rather than left to memory.
The agency conversation about survival figures rather than fill rates.
The buddy scheme with adjusted targets.
And the total cost page, produced and put in front of the budget holders it names.
What the first month should produce
A tenure curve with the spike located.
Twenty exit conversations grouped into causes.
Two visible fixes in place.
A monthly measure and a named owner.
And a costed argument for whatever comes next, which is now evidenced rather than asserted.
If the diagnosis is uncomfortable
Frequently it points at the advert, at a named supervisor, or at a shift pattern somebody designed.
All three are awkward to report and cheap to fix relative to the alternatives.
A diagnosis that avoids them because they are awkward has wasted the month, and everybody who was asked will notice that nothing changed.
The one thing if there is no time
Count last month's starters still present, and ask twenty leavers why they went.
Two days of work, and it is a diagnosis.
Everything else in this collection follows from those two numbers.
What the month costs
Four days of somebody's time, spread across four weeks.
No budget, no system, no supplier.
Which is worth stating plainly when the work is proposed, because the usual objection is resourcing and the usual assumption is that it needs a programme.
How it fails in practice
The month produces a diagnosis, a report and a plan, and nothing anybody on the floor can see.
The next attempt starts from a worse position.
The check
Fix two cheap things in week three, before the analysis is finished.
Visible change is what makes the rest of the work possible.