Planning for an attrition wave: the cohort cliff
Attrition doesn’t arrive smoothly
Capacity plans usually treat attrition as a steady percentage — lose two per cent a month, hire to match, job done. Real attrition is lumpier and more dangerous than that, because people don’t leave at random: they leave on a tenure curve, with a pronounced cliff somewhere in the first year. Hire a big cohort all at once — a peak intake, an acquisition, a class to cover a launch — and you’ve quietly scheduled a synchronised exit months later, when that whole group reaches the leaving cliff together and walks out of a hole in your capacity that a flat attrition assumption never saw coming.
The cliff, and why it surprises people
Early-tenure attrition is front-loaded: a lot of leavers go in the first six to twelve months, as the job turns out to be harder than the brochure, the training-to-floor transition bites, and the ones who were never going to stay self-select out. Plot leavers against tenure and you get a curve that rises to a hump and then falls away as the survivors settle. A single large intake moves through that curve as a block, so its exits cluster — and because the flat “two per cent a month” model spreads those losses evenly across the year, the plan shows a gentle trickle while reality delivers a cliff. The bigger and more concentrated the intake, the sharper the cliff, and the more brutal the capacity gap when it lands.
Seeing it coming, and smoothing it
The defence is to stop modelling attrition as one number and start modelling it by cohort and tenure. Track survival curves for your intakes so you know the shape of your own cliff, and project each cohort forward separately — the synchronised dip appears in the plan months before it appears on the floor, which is exactly when you can still do something about it. Then smooth the input: stagger intakes into smaller, more frequent classes rather than one giant cohort, so the exits spread out instead of stacking up. Watch early-tenure attrition as a leading indicator, because a cohort leaving faster than its predecessors is a warning you can act on. And carry a deliberate buffer for the cliff you can’t avoid. Attrition will always cost you; planning it as the wave it really is, rather than the trickle the spreadsheet prefers, is the difference between a managed gap and a service cliff nobody saw coming.
Pair this with the cost of attrition, what a normal attrition rate is, and new-hire ramp-up.