Tier 1 Is Not One Thing
Tier 1 is not one thing is the proposition that customer-facing productive time, the first tier in The Value of an Hour, is internally heterogeneous in value, and that planning which treats it as homogeneous is blind to its most consequential variable. Inside Tier 1 sit three value bands — high, medium and low — that a single staffing formula sizes as one undifferentiated load. The value bands are distinct from the Three Bands of Work, which sort by how completely work is specified rather than by what it is worth. This page covers the blind spot, the three bands, the illustrative split that shows their shape, and what routing each band to a different handler changes. It does not carry the scoring method; the Value Routing Model owns the four sub-dimensions and the composite value score. Nor does it carry the destination pools and their staffing mathematics; the Three-Pool Architecture owns those. This page argues for classifying Tier 1 at all, and defers the instruments to those pages.
The blind spot
Erlang-lineage planning takes arrivals, handle time and a service target and solves for agents. In that arithmetic a contact is a unit of load, and its worth to the organization is not an input.[1] The blind spot is in practice rather than in theory. The queueing literature has carried priority rules that rank waiting work by its cost of delay since the 1950s, of which the c-μ rule, weighting cost of delay by service rate, is the canonical form.[2] The review literature on call-center operations covers customer segmentation, priority and skills-based routing as established topics.[3] Routine capacity planning uses little of it.[4] Heterogeneity reaches the plan through skills, which sort work by what an agent is competent to do, and Skill-Based Routing executes that sort. Neither the plan nor the router sorts work by what the interaction is worth.
The result is a plan in which every handled minute is equal. The presentation names three symptoms:[5]
- The best agents spend hours on address changes.
- The hardest calls wait in the same queue as the easiest.
- One formula treats them all the same.
The cost of the blind spot is set by the spread of value across contacts, and the spread is wide. A retention contact carries a relationship's future margin; a balance inquiry carries pennies. The service-quality literature has argued that small changes in defection rates move profit disproportionately, because a retained customer's contribution grows over tenure.[6] A formula that weights the two contacts equally is not neutral between them. It under-resources the one that matters, because cheap contacts are more numerous and fill the queue first.
Three value bands
| Band | What it contains | What sets the band |
|---|---|---|
| High value | Retention, complaints, expert judgment; escalated complaints, complex restructuring | What is at risk if the interaction goes badly is the relationship itself and its future margin; judgment is required |
| Medium value | Disputes, billing, payment plans, adjustments | The task is bounded but the resolution needs discretion within rules; errors are recoverable but costly |
| Low value | Balance checks, status inquiries, payments, statement requests; pennies per transaction | The outcome is fully specifiable, the relationship is not at risk, and the customer wants completion rather than attention |
Three properties of the bands matter for planning. First, the band belongs to the interaction type, not the channel; a chat can be high value and a call low value. Second, the band is not a proxy for difficulty. A complex but low-stakes transaction is medium value at most; a short conversation with a customer about to leave is high value. Third, the band is distinct from specifiability. The Three Bands of Work sort work by how completely it is documented and measured, which determines whether a system can take it; the value bands sort by what the work is worth, which determines whether a system should. The two sorts are independent in principle and correlated in practice, since much specified work is also low value, and the cases where they diverge are the cases routing has to get right. A rule-bound payment plan for a customer flagged as likely to defect is specified work of high value; a discretionary goodwill credit on a small account is discretionary work of low value. Routing on either sort alone misplaces both.
The bands are the coarse form of the 1-to-10 value score that the Value Routing Model produces. The Three-Pool Architecture's thresholds, a score of 8 or above for the Specialist pool and 4 or below as a condition for the AA pool, mark off the high and low bands, with the medium band between. All three bands remain Tier 1. Splitting them does not move any work out of doing-the-work; it makes visible that doing-the-work was never one thing.
Breaking down Tier 1: an illustrative split
The presentation gives Tier 1 a worked shape, with each band assigned to a handler class. The shares are illustrative and uncited; they show one plausible shape, and an operation's actual split is produced by scoring its interaction types with the Value Routing Model.[5] The three handler classes are the three pools of the Three-Pool Architecture: human only is the Specialist pool, collaboration the Collab pool, and AI only the AA pool.
| Band | Illustrative share | Routes to |
|---|---|---|
| High value | 15 percent | Human only |
| Medium value | 25 percent | Collaboration |
| Low value | 60 percent | AI only |
Where a split takes this shape, a minority of Tier 1 is value-dense and needs people without qualification, a majority is thin, transactional and fully specifiable, and between them sits a band where people and systems do the work together. Whether an operation's split takes this shape is an empirical question its own scoring answers. The presentation summarizes the operating rule as classify the work, route by readiness: the band determines what the work is worth, and readiness — the system's capability for that interaction type, and the organization's readiness to trust it — determines who takes it. Value alone does not fix the route. A low-value type with low AI capability stays with people; the two-dimensional rule is the subject of Value First, Then Route.
What the split changes
The presentation's reframing is that after the split the organization has not reduced headcount; it has increased the value of every human hour.[5] The claim is narrower than it sounds. The same human hours exist before and after; what changes is their distribution, from spread evenly across three bands to concentrated in the high and medium bands, with the low band carried by systems. The plan's output changes from hours available to value produced per hour. The service-profit chain supplies the causal reading: employee capability at the point of contact is what produces the external value that customers act on.[7] The presentation states the opportunity cost as every hour on low-value work being an hour not spent on retention;[5] the split makes that cost a planning quantity rather than an intuition.
Two consequences follow, and neither is free.
The human queue hardens. Once the low band leaves, the remaining human work is heavier per unit. Handle time rises, the skill floor rises, and targets set on the old mix are missed by construction. This is the ordinary outcome of allocating to machines the tasks that can be specified and leaving people the rest, described for process control long before service automation.[8] The sizing consequences are the subject of The Hardening Residual.
The load changes shape. Two effects run in opposite directions. Moving the low band to systems takes volume out of the human pools, and the Occupancy page's Level 5 position expects AI capacity to absorb demand spikes and protect people from sustained high occupancy. But within the hours that remain, every contact is demanding. In an undifferentiated queue, easy contacts interleave with hard ones and give recovery inside a shift; a queue of only high-band work removes that. The Occupancy page's fatigue mechanism is the absence of recovery between contacts, a function of occupancy level. The split changes what each contact costs rather than how many arrive, so occupancy targets for a high-band pool have to be set on intensity as well as level. Ramp time moves the same way: the Speed to proficiency curve gives typical ramp times of nine to eighteen months for complex specialties against three to six for routine work. The split does not remove these costs; it makes them visible and plannable.
A practitioner reading
An operation can test the proposition on its own data before adopting any model.
- Inventory the interaction types, at the level the routing platform already distinguishes.
- Assign each type to a band using the three properties above, without scoring precision at this stage.
- Measure handled hours by band, and by agent tenure cohort within band.
- Find the hours the most experienced cohort spends in the low band. That figure is the blind spot, in hours.
- Estimate the value of those hours redeployed to the high band, using retention and complaint outcomes as the yardstick.
Where the figure in step four is large, and it usually is because low-band work is numerous and arrives first, the case for the split is made in the operation's own hours. The exercise then shows why the split cannot be executed as a queue setting. Moving the low band to systems, and the high band to the people best placed for it, is a routing design, and it is the subject of Value First, Then Route.
Maturity Model Position
Under the WFM Labs Maturity Model™, the blind spot is invisible at Levels 1 to 3, where the unit of planning is the contact and heterogeneity is handled through skills. The proposition becomes operational at Level 4, where types are classified by value before routing and the three pools exist to receive the bands. At Level 5 the bands are re-sorted continuously as system capability changes the readiness dimension.
See Also
- The Value of an Hour — the four-tier taxonomy in which Tier 1 sits
- Value First, Then Route — the routing mechanic that consumes the bands; why value alone does not fix the route
- Value Routing Model — the scoring method that produces an operation's actual split
- Three-Pool Architecture — the destination pools and their staffing mathematics
- Three Bands of Work — the independent sort by specifiability
- Skill-Based Routing — the sort by competence that the value sort complements
- The Hardening Residual — what happens to the human queue once the low band leaves
- Occupancy — why the split moves occupancy level and contact intensity in opposite directions
- Planning for Agentic AI Is Not a Carve-Out — why the low band does not simply disappear from the plan
- Value-Based Planning Model — the model the split serves
References
- ↑ Gans, N., Koole, G., & Mandelbaum, A. (2003). "Telephone Call Centers: Tutorial, Review, and Research Prospects". Manufacturing & Service Operations Management 5 (2), 79–141. doi:10.1287/msom.5.2.79.16071.
- ↑ Cobham, A. (1954). "Priority Assignment in Waiting Line Problems". Journal of the Operations Research Society of America 2 (1), 70–76. doi:10.1287/opre.2.1.70.
- ↑ Akşin, Z., Armony, M., & Mehrotra, V. (2007). "The Modern Call Center: A Multi-Disciplinary Perspective on Operations Management Research". Production and Operations Management 16 (6), 665–688. doi:10.1111/j.1937-5956.2007.tb00288.x.
- ↑ Practitioner observation across contact-center capacity plans; a consistent pattern rather than a measured result.
- ↑ 5.0 5.1 5.2 5.3 Lango, T. (2026). Adaptive: Building Workforce Systems for an (Unpredictable) Future. Presentation, SWPP Annual Conference, slides 37–38. The three bands, the illustrative 15/25/60 split and the handler assignments are the presentation's framing; no source is cited on the slides for the percentages.
- ↑ Reichheld, F. F., & Sasser, W. E. (1990). "Zero Defections: Quality Comes to Services". Harvard Business Review 68 (5), 105–111.
- ↑ Heskett, J. L., Jones, T. O., Loveman, G. W., Sasser, W. E., & Schlesinger, L. A. (1994). "Putting the Service-Profit Chain to Work". Harvard Business Review 72 (2), 164–174.
- ↑ Bainbridge, L. (1983). "Ironies of Automation". Automatica 19 (6), 775–779. doi:10.1016/0005-1098(83)90046-8.
