Shrinkage: the planner’s hardest input

Forecasting · ~7 minute read

Why shrinkage is harder than it looks

Ask three workforce planners what their operation’s shrinkage is and you will get three different numbers. Ask the same three planners how they measured it and you will get three different methodologies. Ask any of them why their schedules keep coming up short and the most common answer, eventually, is that shrinkage was higher than the plan assumed. Of the three inputs that drive a contact centre forecast — volume, AHT, and shrinkage — shrinkage is the one that hides the most error, generates the most disputes, and pays back the most to a planner who learns to handle it properly. This article walks through what shrinkage really is, why every operation measures it slightly differently, the three sources of error, and how to track and reduce it deliberately.

What shrinkage actually is

Shrinkage is the percentage of paid agent time that is not spent answering customers. Everything that takes an agent away from a contact — annual leave, sickness, training, coaching, meetings, breaks, system downtime, queue idleness counted as planned time, and any number of smaller categories — aggregates into the shrinkage number. The planner uses that number to convert net agent-hours required (the output of an Erlang calculation) into gross paid hours and headcount. The relationship is the simplest piece of arithmetic in the planner’s job: gross = net / (1 − shrinkage). The relationship is also where most operations silently fail, because the shrinkage number that goes into the formula is wrong.

The compound trap

The most common error in shrinkage modelling is adding the components together. Annual leave is 11%, sickness is 5%, training is 3%, breaks are 8%, meetings are 2% — total 29%. This is wrong, and the error grows with the number of components. An agent on annual leave cannot also be in training. The components are not additive but multiplicative: net productive time is (1 − a)(1 − b)(1 − c)… through every component, and total shrinkage is one minus that product. The same component breakdown applied multiplicatively gives a shrinkage of around 26%, not 29%. The error is small in absolute terms but consistently wrong in the same direction, and over a year of schedules it adds up to real money. The shrinkage calculator on this site handles the compounding automatically and shows the difference between the naive and correct sums.

Why every operation measures it differently

There is no single industry definition of shrinkage. Some operations include public holidays in the baseline contract hours and exclude them from shrinkage; others put them in. Some count paid breaks as productive time on the grounds that they are short and statutory; others count them as shrinkage. Some include sick pay; some include only the agent’s absence, regardless of whether it was paid. Some include training delivered by the agent themselves; some do not. The result is that a quoted shrinkage of 30% from one operation may be the same real-world situation as a quoted 22% from another, simply because the boundary of the term has been drawn differently.

This is more than a definitional curiosity. It matters every time the planning team benchmarks against another operation, every time finance compares the operation against budget assumptions inherited from a previous regime, and every time a senior manager asks why the same headcount is producing different productive output than it used to. The planner’s first job, when arriving in a new operation or when shrinkage suddenly becomes a topic, is to write down the operation’s definition explicitly. Without the definition, every conversation about shrinkage talks past itself.

Three sources of error

Beyond the additive-vs-multiplicative trap and the definition trap, three substantive sources of shrinkage error recur. The first is under-counted absence. Sickness, particularly intermittent short-duration sickness, is consistently lower in the assumption than in reality because the planner is looking at the official absence return rather than the actual coverage gap. An agent who is in but on light duties, or in but doing development time, or in but tied up in an investigation, is not on the phones — even if the absence report says they are at work.

The second is creeping training time. Operations under-budget the training, coaching, and quality time agents actually receive. A planner who assumes two percent for coaching when the operation is actually delivering five percent will under-staff predictably. The fix is to measure what the operation does, not what the standard recipe says it does.

The third is queue-idle reclassification. When volume is low and agents are sitting in available state with no calls, some operations record this as productive time (they were ready to take calls), some as shrinkage (they were not handling calls), and some as a third category that gets shifted between the two depending on the conversation. The categorisation choice is methodological, but it must be consistent. Inconsistency is the source of the “our shrinkage seems to move whenever volume moves” complaint.

How to track shrinkage well

The basics of good shrinkage tracking are unglamorous and effective. Decompose the total into the same set of components every time you measure it. Measure each component in actual hours, not estimated percentages, against actual paid hours. Track it weekly, with rolling four- and twelve-week averages alongside the weekly number so noise does not provoke over-reaction. Compare it to budget every cycle and explain any variance by component, not as a single blended number. Publish the methodology so any analyst in the team produces the same result from the same data.

Two further habits separate competent shrinkage tracking from excellent shrinkage tracking. The first is tracking shrinkage by team and by tenure. Different teams and different tenure bands have systematically different shrinkage profiles — new hires take more training time, older agents have higher leave entitlement, certain teams attract more coaching. Aggregating to a single operation-level number hides the variation that drives the staffing accuracy of specific queues. The second is forecasting shrinkage separately. Most operations forecast volume and AHT carefully and assume shrinkage will be whatever it was last year. Shrinkage has its own seasonality — bank-holiday weeks, summer holiday weeks, post-Christmas absence patterns — that deserve their own model.

Reducing shrinkage vs. forecasting it accurately

These are different jobs, and operations sometimes confuse them. Forecasting shrinkage accurately means the schedule reflects what will actually happen, which is good for service level and cost discipline. Reducing shrinkage means changing operational practice to drive the number down, which is good for productive capacity and unit cost. Both are valuable, but only one belongs to the planner. The planner’s job is the forecast. The operations leadership’s job is the reduction. A planning team that quietly assumes a lower shrinkage than reality, hoping operations will catch up, ends up under-staffing predictably and then being blamed for the consequences. A planning team that forecasts what will actually happen and then helps operations see where the reduction opportunities are stays credible and useful.

Common mistakes

Five patterns recur. Adding components instead of compounding them, as discussed above. Using last year’s annual average and assuming next week will match it, which ignores week-on-week variation that often dwarfs the annual move. Treating shrinkage as a single number rather than a structured breakdown. Forecasting shrinkage that the operation has been told to deliver rather than the shrinkage the operation actually produces — an honest but politically uncomfortable distinction. And, finally, ignoring the difference between paid shrinkage and unpaid absence, which matters when the conversation moves from operational coverage to actual cost.

Conclusion

Shrinkage is unglamorous and decisive. It is the input that breaks more schedules than any other, the topic that creates the most friction between planning and operations, and the number that finance and senior management trust the least when it surfaces in a variance conversation. The planners who handle shrinkage well decompose it, measure it consistently, forecast it deliberately, and report it honestly. The operations that get the planning right invest in the conversation about reducing it, separately from the conversation about modelling it. Get those habits in place and shrinkage stops being the surprise that quietly ruins the quarter; ignore them and it remains exactly that.

Use the shrinkage calculator on this site to model the compound build-up of your own components and convert net agent requirement to gross headcount. Pair this article with the beginner’s guide to forecasting for the wider context.