The shrinkage assumption that’s always wrong
Modelled 30%, realised 36%
The schedule was built assuming 30% shrinkage. The actual is 36%. The six-point gap is six points of headcount that doesn’t show up where the operation needs it, and the operation can’t hit SL because the effective capacity is six percent lower than the schedule said it would be. The dashboard reports schedule on plan and SL off plan, and nobody can quite explain why. The diagnosis often lands on “poor adherence” or “higher absence,” both of which deflect from the actual cause: the shrinkage model captured the planned categories and missed the unplanned, small, cumulative ones.
Where the gap lives
Five sources, each individually small, cumulatively large.
Ad-hoc coaching overruns. A planned 30-minute coaching session that runs 45 minutes. Multiplied across a team of 15 agents, once a week, this is 3.75 hours a week of unplanned shrinkage no model captures.
TL-pulled-into-escalation. The TL needed for an escalation, taking an agent off the floor for the briefing. Operations escalations average 30–45 minutes; if they happen twice a week, that’s another 1–1.5 hours per agent per week not modelled.
System glitches. Five-minute log-outs at the start of the shift, mid-shift connectivity issues, the call-recording system that needs restarting. Each is small; cumulatively across the year they add 1–2 percent of shrinkage.
The exit-interview week. When an agent gives notice, their final two weeks involve hand-overs, exit interviews, HR conversations, and reduced focus. Operations with 30% annual attrition lose a meaningful chunk of effective capacity to this layer alone.
Adherence drift. The agent who’s 3 minutes late back from each break adds 12 minutes a day of un-modelled aux time. Across the floor, this is the single largest source of unmeasured shrinkage in most operations.
How to measure realised vs modelled honestly
The exercise is straightforward but operations rarely run it. Take six weeks of historical data. Compute total scheduled hours, then total productive hours (talk + ACW), then everything in between. The shrinkage figure is one minus (productive / scheduled). Compare to the modelled shrinkage that the schedule was built against. The gap is usually 4–8 percentage points in operations that haven’t looked.
Then break the gap down. Categorise the un-modelled time using whatever aux codes the operation has. The biggest single contributor will surprise the planning team. It usually isn’t the one operations leadership thinks it is.
How operations close the gap
Three approaches that work.
Add a “cumulative unplanned” layer to the shrinkage model. The honest answer is that the schedule should be built against 36%, not 30%, in the example above. Adding a 5–8 percent allowance for the cumulative unplanned matches reality and produces a schedule the operation can actually deliver.
Drive the underlying behaviour. Adherence is the largest single component for most operations. Investment in adherence reporting, TL accountability, and tight-loop coaching typically claws back 2–3 points of shrinkage at no cost. See adherence and conformance.
Restructure the unplanned categories. Ad-hoc coaching, escalation support, and system-glitch time should be planned categories, not unplanned residuals. If coaching consistently takes 45 minutes rather than 30, the model should plan for 45.
The honest commercial case
The shrinkage gap has a direct cost. Six points of shrinkage on a 100-agent operation is six agents of capacity that the operation paid for and didn’t get. At a fully-loaded cost of £35k per agent, that’s £210k of payroll producing no service. Operations leadership tends to find this number more motivating than the abstract “we have a shrinkage gap.”
Conclusion
The shrinkage assumption is the input the schedule is most sensitive to and the input most operations get wrong. Closing the gap doesn’t require new methodology; it requires honest measurement of realised vs modelled, a willingness to update the schedule against the higher (real) number, and investment in the underlying drivers. Operations that do this find their schedules become deliverable; operations that don’t spend years explaining why a schedule that looks right on paper produces SL that’s consistently off.
Next in the series: The multi-skill illusion.
Pair this with shrinkage deep-dive, adherence and conformance, and the cost of perfect adherence.