The Value of an Hour

Value per handled hour by tier. Some hours create value; some destroy it once the customer's time is counted. The automation zone is defined by sign, not by cost.

The value of an hour is a planning lens that asks what each paid hour of a service employee's time returns, rather than whether that hour was available to handle contacts. It sorts paid time into four tiers by the value each produces: doing the work, getting ready, the cost of being human, and the leak. The lens supplies the supply-side value term in value-based planning: what an hour of the workforce is worth. The Value-Based Planning Model supplies the demand-side term, the value of each interaction, through its interaction taxonomy. It divides the work with two metric pages, Shrinkage and Occupancy, which measure the availability and use of paid time and are not restated here. This page values that time, and shows why the reading of both metrics changes once tiers replace the available-or-unavailable binary.

Three founding questions

The presentation from which the taxonomy is drawn enters value-based planning through three questions, asked in a fixed order:

1. What is the value of every hour invested in employees?
2. What is each interaction worth from the customer's perspective?
3. How should work be routed to maximize value, not only to minimize cost?

The order matters: a route assigns an hour of someone's time to an interaction of some worth, so the third question needs answers to the first two. The first question is this page; the second is taken up by Tier 1 Is Not One Thing and scored by the Value Routing Model; the third is the mechanic in Value First, Then Route.

The first question departs from how planning usually treats time. Erlang-lineage staffing takes handle time and availability as inputs and solves for the people needed to meet a service target. In that arithmetic an hour is either available or it is not, and its cost is the same whatever it produces. The marketing-science treatment of service productivity argues that productivity is a decision variable to be set at the level that maximizes return, because above that level the revenue lost to poorer service exceeds the cost saved. If hours differ in what they return, they have to be counted differently before that level can be found.

The four tiers

Tier	Name	Definition	Examples	Value rating
1	Doing the work	Customer-facing productive time	Calls, chats, claims processing, back-office fulfillment	High
2	Getting ready	Enabling work that builds capability	Training, coaching, development, team meetings	Positive
3	The cost of being human	Necessary; neither positive nor negative	Breaks, lunches, paid time off, vacation	Neutral
4	The leak	Cost with no output	Unaccountable time, idle without assignment	Negative

Tier 1, doing the work, is the time the operation exists to produce: handling contacts and completing the work behind them. It is the tier queueing models size and the only tier in the numerator of occupancy. It is rated high but is not uniform; the split inside it is the subject of Tier 1 Is Not One Thing.

Tier 2, getting ready, is time spent building the capability Tier 1 will draw on. It is rated positive rather than high because its return is deferred: a coaching hour produces nothing today and a better hour next week. The service-profit chain places this tier at the start of its causal sequence: internal service quality drives employee satisfaction, retention and productivity, which drive the external service value that customers experience. Treating it as lost availability discards that sequence.

Tier 3, the cost of being human, is time that people must take and that returns nothing directly: breaks, meals, leave. It is neutral by construction, a condition of employing people at all. The planning error it invites is the attempt to reduce it below what recovery and regulation require.

Tier 4, the leak, is paid time with no output and no assignment: unaccountable intervals, idle time with nothing scheduled into it. It is the only tier rated negative. It has to be distinguished from the idle time that queueing requires. A queue that meets a service target must carry slack, because an agent who is always busy is an agent no arriving customer can reach; the size of that slack is set by the service-level target and the pool's scale. That required slack is not Tier 1 content, since no work is done in it, but Tier 1 carries it: staffed-to-queue time the service target has already priced. The leak is the idle beyond it, which nobody planned and nobody assigned.

At the desk the two look identical: a ready state and no contact. They are separated at the interval. The service-level calculation for an interval implies a planned occupancy, and therefore a planned quantity of idle; idle in excess of that quantity is leak. Variance Harvesting uses the same planned-against-actual occupancy decomposition to tell a genuine surplus from the compensating movement of occupancy; the leak is that surplus left unassigned. The leak also straddles the two metrics: its unassigned-ready part sits inside occupancy's denominator, its unproductive-auxiliary and system-issue parts inside shrinkage. No single metric shows the whole of it.

Reading shrinkage and occupancy through the tiers

The metric pages decompose paid time by cause but score every component identically in the staffing arithmetic. Shrinkage sums every hour in which an agent cannot take a contact, so a training hour, a lunch and an unexplained absence all reduce availability by the same amount. Occupancy then asks what fraction of the available hours was spent handling. Both metrics are correct as measures. What the four tiers change is how they are read.

Read through the tiers, shrinkage is a mixture of Tier 2, Tier 3 and part of Tier 4, and the three behave differently under pressure. Cutting Tier 2 improves the shrinkage figure and destroys future capability. Cutting Tier 3 improves the figure and lowers measured occupancy, because more time becomes available at the same volume, while removing the recovery that breaks and leave exist to provide; the plan looks less busy and the people are more tired. Cutting Tier 4 improves the figure and costs nothing. A shrinkage target cannot tell these apart, and a plan that pursues it will cut whichever is easiest. The tier view replaces the target with two: drive Tier 4 toward zero, and budget Tier 2 as an investment with a return.

Read through the tiers, occupancy has two limits. Its numerator treats every handled minute as equal, when Tier 1 contains work of very different worth. Its denominator, available time, contains the leak: an agent in a ready state with nothing assigned counts as available, so a pool with a large leak reports low occupancy and looks overstaffed. The Occupancy page's own section on agent state discipline makes the same point from the measurement side: what occupancy measures depends on which states agents signal, not on what they are doing. Neither limit makes occupancy wrong. Both mean it cannot answer the first founding question on its own.

The practical link to existing practice is Variance Harvesting. One of its moves, capturing an in-day surplus in a demand dip and assigning it to coaching or training, converts Tier 4 into Tier 2. Its protected-break response converts surplus into Tier 3 instead, and voluntary time off removes the hour from paid time altogether. The taxonomy names what each conversion is worth.

The automation zone

Automation acts on the boundary between Tier 1 and Tier 2. On the Tier 1 side, agentic systems take the codifiable, transactional handling: the tasks that can be fully specified are the tasks that machines substitute for, while the remainder is complemented rather than replaced. On the Tier 2 side, the same systems deliver training, surface knowledge during a contact and shorten a new hire's time to proficiency.

Automation therefore changes the composition of an hour, not only the count of hours. An hour freed from Tier 1 flows to Tier 2 if the plan assigns it there, or to Tier 4 if it does not. The carve-out habit described in Planning for Agentic AI Is Not a Carve-Out assumes the hour is gone; the tier view insists it lands somewhere, and that where it lands is a planning decision. What stays in Tier 1 after automation is harder per unit than before, the subject of The Hardening Residual.

Reading a plan through the tiers

A planning function can adopt the lens without new tooling.

• Tag every schedule state to a tier. The mapping runs over states the platform already carries, and it exposes how much of reported shrinkage is Tier 2 and how much is Tier 4.
• Report hours by tier alongside shrinkage and occupancy. The metrics remain; the tier report explains them.
• Set the leak, not shrinkage, as the reduction target. Tier 3 is fixed by policy and regulation; Tier 2 is a budget line with a return; only Tier 4 is waste.
• Pre-assign the destination of freed hours. When automation takes Tier 1 work, the plan should say in advance which tier receives the released hours.
• Diagnose low occupancy before acting on it. Compare the interval's idle against the idle its service target implies. The implied portion is priced slack; the excess is surplus, and it is leak only where it went unassigned.

Maturity Model Position

Under the WFM Labs Maturity Model™, Level 1 manages shrinkage as a single figure. Level 2 tracks it by component and separates planned from unplanned, but every component still feeds one staffing divisor with the same weight, so the tier view is available only as a reporting overlay. At Level 3, Variance Harvesting begins converting the leak into enabling time in-day, and the taxonomy becomes an operating distinction. The lens is a Level 4 concept: Level 4 is the first level at which the value of an hour is an input to the plan rather than a commentary on it. Value classification of Tier 1 and pool-specific staffing both depend on it. At Level 5 the destination of freed hours is decided continuously as capability moves.

See Also

• Shrinkage — the availability metric this page re-reads
• Occupancy — the utilization metric this page re-reads; its agent-state-discipline section
• Tier 1 Is Not One Thing — the three value bands inside doing-the-work
• Value First, Then Route — the routing mechanic that consumes the value of an hour
• Planning for Agentic AI Is Not a Carve-Out — why freed hours land somewhere rather than disappearing
• Value-Based Planning Model — the demand-side term: interaction value through the interaction taxonomy
• Variance Harvesting — the in-day conversion of leak into enabling time
• Coaching and Agent Development — the largest Tier 2 component
• Deferrable Work and the Idle Cushion — staffed-to-queue slack as priced time rather than leak
• Adherence and Conformance — how the states that feed the tiers are measured
• The Hardening Residual — what happens to the Tier 1 that remains after automation