The Headcount Gap Hiding Between Your Multi-Location Sites

For contact centers planning across multiple locations, here’s a practical check on whether each site's plan reflects its actual operating conditions.

When was the last time you looked at your shrinkage assumption? Not at the network level, but for each site?

Most multi-location contact center workforce management runs to a single blended assumption. It's practical. It's faster. And for a long time, it's been accepted as good enough. But good enough at the network level and good enough at the site level are two different things. The gap between them doesn't show up in your plan, it shows up in execution.

How the Gap Forms

When multi-location workforce planning applies a network average to multiple sites, you're making a specific claim: that every location behaves similarly enough that a single assumption is an accurate representation of all of them.

Sometimes that's roughly true. More often, it isn't.

Take shrinkage. In a contact center with two operating sites (one larger site with a primarily in-house workforce, one with a different staffing model) it's not unusual for actual shrinkage to differ by 15 percentage points or more between locations. The in-house site might run at 30-40% once you account for all the ways time leaves the floor. The other might run at 20-25%, with a tighter operating model and fewer hours lost to the same factors.

Weighted by volume, the blended average comes out to roughly 29%. A reasonable number, but an inaccurate one for both sites.

Shrinkage is one variable, but it's not the only one. Average Handle Time (AHT), new hire learning curve, attrition – all of these can differ meaningfully between sites while a network plan treats them as equal. Each discrepancy is small in isolation. Combined, they determine whether the staffing allocated to each site reflects what that site actually needs to meet the expected work demand.

What the Gap Actually Means

When these variables diverge between sites and your network-level assumption, the allocation each site receives no longer reflects what it requires.

A site running higher shrinkage (or longer AHT, or a larger share of newer agents) has less productive capacity than the plan assumes. Coverage gets thin before peak. Service level slips. Meanwhile, another site may be carrying capacity it doesn't need, running occupancy low without anyone asking why.

What typically shows up in your end-of-period review isn't "the assumption was wrong." It's "Site A missed its capacity requirements." Attribution: demand variability. Forecast error. The plan doesn't get questioned, because at the network level, the variances partially cancel. The aggregate looks “close enough”.

That's the misattribution problem. The error lives in the planning assumption. The blame lands on the forecast.

There's also a structural reason this repeats: multi-location workforce planning at the network level, even with carefully constructed weighted averages, produces an aggregate view. It doesn't surface what each site actually requires. A plan that's net-accurate can still be wrong at every individual location.

What to Check in Your Multi-Location Workforce Plan

Four variables most commonly drive a gap between what a network plan allocates to a site and what that site needs:

Shrinkage. Often applied as a single blended rate. A site running 10-15 percentage points above or below your planning assumption (typical in-house vs. BPO variance) is receiving meaningfully different staffing than its operating reality requires.

Average Handle Time. A site handling a different contact mix, or one with less tenured agents, may have AHT that diverges from your planning assumption. Longer calls need more coverage, and the plan won't reflect it if the AHT is averaged across all sites.

New hire learning curve. If a site is growing or experiencing elevated turnover, a portion of its headcount is operating at partial productivity. That reduces effective capacity in ways that don't appear in a standard shrinkage figure.

Attrition. High-attrition sites cycle through the early part of the learning curve continuously. Headcount may be maintained by just in time hiring and training, but productive capacity is structurally lower than a stable-tenure site of the same size.

Pull your actual figures for these four variables at each site and compare them against your current network planning assumption. Where the spread is widest is where your plan is carrying the most unexamined risk, independent of anything happening with your demand forecast. 

Beyond planning accuracy, there's a second benefit to knowing where your assumptions diverge: it gives you somewhere to look for improvement. A site with a longer new hire learning curve isn't just a planning input, it's a signal. When you can compare outcomes across locations side by side, you can start asking why one site performs differently from another, and whether what's working in one place can be applied elsewhere. That's a different conversation than reviewing aggregate numbers at the end of the period.

If the gap is small across the board, that's useful to know. If it isn't, you now have something concrete to bring into a conversation about where the plan actually stands.

How closely do your site-level operating conditions match what's currently in your network plan? And the last time a specific location missed its headcount requirements, do you know whether the issue was the forecast, or the assumption underneath it?

Cinareo is a capacity planning platform built for contact centers. If you're working through multi-location planning challenges and want to talk through how others are approaching them, we're glad to have the conversation.