Decision Authority Tiers in Workforce Operations
Decision Authority Tiers in Workforce Operations is the convention that fixes the authority tier field of every generic process card: a five-rung ladder of seats, from the analyst acting inside published guardrails to the executive sponsor acting by exception, at which each decision point of a process is signed. It matters because a function that has not written down which seat may take which decision either freezes (every lever waits for a manager) or drifts (every analyst sets a different threshold), and in both cases no automation rule can be released, because there is no stated tier to release it into. The convention is part of Process Standardization and fills the authority tier row of the process profile on Process Decomposition (L0–L3).
The tiers name positions on a ladder, not job titles: tier 1 is the real-time analyst in the real-time pillar, the forecaster inside the lock calendar in forecasting, the scheduler inside the published rules in scheduling, and what makes a seat tier 1 is that its decisions are bounded by guardrails a higher seat published. The ladder follows the principle that a right sits with the seat that holds the knowledge the decision needs, and that where the knowledge is specific to a shift, a queue or a client portfolio it is cheaper to move the right than the knowledge;[1] the working rule that a decision sits as low as the seat whose error is cheapest to reverse is practice [A]. One accountable seat per decision is the first condition the decision-roles literature sets for a decision that holds.[2]
The five tiers
| Tier | Seat | What it signs | What bounds it |
|---|---|---|---|
| 1 | the pillar's analyst: real-time analyst, forecaster, scheduler, the desk that runs loans and trades | actions inside published guardrails, from an approved cascade; no approval sought | the guardrails, held as SET LOCALLY values in the L3 job aids |
| 2 | supervisor or team lead | actions the guardrails exclude by definition, on the analyst's recommendation; minor changes to L2 and L3 | the pillar's budget for the day or week |
| 3 | operations manager or ROC duty lead | actions that cross a queue, skill, site or node; incident ownership at the top severity rows; a reforecast outside the band the schedule can absorb | the plan of record; the decision-rights table |
| 4 | function head | the mid-term and long-range locks, the plan of record, the build parameters, a statutory variant, a version of the standard | the executive sponsor's envelope; the statute |
| Exception | executive sponsor | a gap outside the band the plan can absorb; one client's service traded against another's; a change to the standard's scope | by exception only; every exception recorded with its price |
A seat may always take a lower tier's decisions; it may never take a higher tier's. An analyst who takes a tier 2 action because the supervisor was unreachable has taken an unsigned action, and the record says so.
Variants map to the tiers as Standard and Variant in Process Documentation assigns their approvers: statutory, the function head at tier 4 on legal advice; contractual, the placement and vendor lead, a tier 3 seat of the placement pillar; structural, the process owner at whatever tier that seat holds; transitional, the standards committee, which sits outside the ladder because it governs the standard, not the operation.
Tiers, colour codes and severity
Three instruments describe a real-time situation and are commonly merged; each answers a different question.
The staffing colour codes classify what the operation has become; Incident Management for Contact Centers carries them as the escalation-code job aid, Blue through Black. Typically [A] Blue is a surplus the desk may release, Green a position inside its band, Yellow outside the band with levers available, Red outside the band with first-tier levers exhausted and a ticket required, Black a loss of service or control; triggers and actions per code are SET LOCALLY. Each code maps to a tier, written once in the job aid: Blue and Green held at tier 1; Yellow acted on at tier 1 with tier 2 informed; Red at tier 2 with tier 3 owning the ticket; Black run by the tier 3 duty lead with tier 4 informed.
The severity matrix on Contact-Center Incident Severity Matrix classifies what happened to the operation, values set locally. Severity decides who is told and how fast; the tier decides who may act. They meet at the top rows: an incident at severity 1 or 2 fires the continuity wire and is owned at tier 3, with the tier 4 seat among those notified. A severe incident under control at tier 3 and a Black state with no single incident behind it are representable only because the instruments are separate. The tier is a sign-off, not a state: it is read from the card's profile, not from the wallboard.
Tiers and the action gate
Human Gates and Number Grades places an action gate in front of every platform change a real-time team could make, and Real-Time Agents states the rule: inside guardrails a real-time analyst approves, outside them a manager; the agents propose without executing. The tiers are what "inside" and "outside" refer to. A rule is written against a class of action, the class carries the tier of the seat whose decision it replaces, and the rule is released into that tier, never above it: a rule that proposes break moves is released into tier 1 when its catch rate for that class is published and stays a proposal above it whatever the catch rate, because evidence about break moves says nothing about mandatory overtime. The staging is the levels-of-automation one, authority extended a stage at a time as reliability is measured.[3] The tier bounds what a released rule may touch and names the seat whose verification work instruction produces the catch rate (Runbooks, Playbooks and Work Instructions).
Tiers and decision rights by horizon
Decision Rights by Planning Horizon fixes the first rule: the function is accountable for the method and responsible for the advice; the node owner is accountable for the decision. The tiers do not reach past it. A tier names the seat inside the function that signs the function's method or advice: the forecaster signs the signal applied, the function head signs the lock and the gate's outcome as advice. Where the RACI makes the node owner accountable, a hiring plan, a placement, a clearing move, the decision is a RACI row, not a tier, and the card's profile reads "advice at tier n; decision per the decision-rights table". The exception tier is the route that page's forums use when a gap exceeds the band the plan can absorb, not a sixth decision body.
Typical decisions by tier
Illustrative, graded [A]; every cell is a value a function sets on its own cards.
| Pillar | Tier 1 | Tier 2 | Tier 3 | Tier 4 | Exception |
|---|---|---|---|---|---|
| Forecasting | apply a confirmed, sized signal inside the lock calendar | a judgmental override outside the model's band, reason recorded | a reforecast that moves the weekly requirement outside the schedule's band; the short-term lock, signed by the planner who owns the client portfolio | the mid-term and long-range locks; the plan of record | a gap outside the band the plan can absorb |
| Scheduling | swaps, time-off inside the allotment, break moves inside the rules | a time-off exception beyond the allotment; a shift change inside the notice period | re-publication after the lock; the week's overtime plan | the build parameters; the allotment plan | a collective-agreement or works-council change |
| Real-time | break resequencing, skill rebalancing inside guardrails, voluntary time off inside budget | mandatory overtime; pulling agents from off-phone activity | cross-site or cross-skill routing change; incident ownership at severity 1 or 2; partner overflow | suspending a service construct for the day; a standing guardrail change | one client's service traded against another's |
| Work placement | a loan inside pre-agreed terms | a trade between internal pools inside the day's cap | a sourcing recommendation through the gate, as advice; a contractual variant | the gate's outcome as advice; the scorecard findings | a placement that changes the network design |
| Continuity interface | receive the wire's notification; run the degradation runbook's first steps | confirm the crisis staffing state | the crisis re-plan; placement under duress proposed | the re-plan signed as the product to the continuity owner; the judgmental override | moving work across nodes where the wire's authority was left unsettled |
| Cross-cutting | publish a report on its calendar with its grades | a minor change to L2 or L3 | a major change request admitted to the queue | a version of the standard; closure of a compliance finding | a finding with statutory exposure |
The judgmental override appears at tier 2 under forecasting and at tier 4 under the continuity interface; the difference is crisis. Inside the normal cycle an override is a bounded correction a team lead can sign; under the wire it is the input to the re-plan and is signed where the re-plan is (Judgmental Forecasting in Crisis (generic process)).
Escalation and step-down
The real-time escalation ladder, analyst to supervisor to operations manager with its triggers and response windows, is on Real Time Threshold Alerts and Escalation Protocols and is not restated; this page generalizes the triggers to every pillar and adds the rule for coming back down.
A decision escalates one tier when any of three conditions holds: the action taken has not produced its expected effect inside the window SET LOCALLY for it; the action now required is above the current tier by definition; or situations compound so that the tier's scope no longer covers the whole. Escalation is one tier at a time except where the card's profile names a direct route, as the severity matrix does for its top rows; incident-handling guidance outside the contact center uses comparable criteria.[4]
A step-down needs the impact confirmed neutral, the reverse check. Before a situation is handed back down a tier, the seat holding it confirms and records three things: the triggering metric has recovered and held for the window set locally, not merely crossed back once; reversing the levers deployed (restoring breaks, releasing overtime, returning borrowed agents) is projected not to reopen the gap, a reforecast question rather than a hope; and no dependency remains open, a partner activation inside its notice period, a pending reforecast, an unclosed incident record. A step-down that fails any part is a new escalation, not a retry. Step-down is one tier at a time, the seat that escalated is told, and the record carries both the escalation and the return, which lets the weekly review (Variance Harvesting) read how long each tier held the situation. Crisis Trigger and Re-Plan (generic process) runs the same check before the crisis re-plan is retired.
The asymmetry is deliberate: escalation is fast because delay costs service; step-down is evidenced because a premature return costs a second escalation, and the human-factors literature documents complacency and overreliance in supervisors of reliable automation unless the return requires them to produce a finding.[5]
Maturity Model Position
Four scales on this wiki use the word level; the launch page states which is which. On the WFM Labs Maturity Model™ a written tier per decision point, with the colour-code mapping in a job aid, is Level 2 practice. The tier as the release target of an automation rule is the Level 3 form; at Level 4 it is a field the placement engine and the real-time agents read, so a proposal arrives addressed to the seat that may sign it.
See Also
- Process Standardization — the hub this convention belongs to
- Process Decomposition (L0–L3) — the process profile whose authority-tier row this page fills
- Decision Rights by Planning Horizon — the first rule and the RACI the tiers sit beneath
- Real Time Threshold Alerts and Escalation Protocols — the real-time escalation ladder and its triggers
- Incident Management for Contact Centers — the escalation codes, Blue through Black
- Contact-Center Incident Severity Matrix — severity, kept separate from the tier
- Human Gates and Number Grades — the action gate and the release condition
- Standard and Variant in Process Documentation — the variant classes and their approvers
References
- ↑ Jensen, M. C., & Meckling, W. H. (1992). "Specific and General Knowledge, and Organizational Structure". In Werin, L., & Wijkander, H. (eds.), Contract Economics. Blackwell, Oxford, 251–274.
- ↑ Rogers, P., & Blenko, M. W. (2006). "Who Has the D? How Clear Decision Roles Enhance Organizational Performance". Harvard Business Review 84(1), 52–61. hbr.org.
- ↑ Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). "A Model for Types and Levels of Human Interaction with Automation". IEEE Transactions on Systems, Man, and Cybernetics — Part A 30(3), 286–297.
- ↑ Cichonski, P., Millar, T., Grance, T., & Scarfone, K. (2012). Computer Security Incident Handling Guide, NIST Special Publication 800-61 Revision 2. National Institute of Standards and Technology, section 3.2.6. nvlpubs.nist.gov.
- ↑ Parasuraman, R., & Riley, V. (1997). "Humans and Automation: Use, Misuse, Disuse, Abuse". Human Factors 39(2), 230–253.
