Production Accounting for Blended Agents
Production accounting for blended agents is the practice of attributing each unit of production and each minute of handling time on an agent who works more than one channel to the channel that produced it, so that no email, case or chat item is booked as idle time on the voice platform. It matters because a blended agent on a mixed estate typically answers email while sitting in a voice available or auxiliary state, the telephony platform records that time as idle availability, and the error reaches both channels: voice occupancy falls, the idle cushion the real-time function believes it can draw on grows, and the email requirement is computed on handling time no system captured. The practice is part of the Workforce Data Definitions Framework and sits between Agent Time Accounting Model, where every minute lands in exactly one bucket, and Multichannel Capacity Model Assembly, which builds one requirement per channel from those buckets.
The mechanism follows from the telephony-centred definition of availability: an agent is available when in the phone system and ready for a contact, auxiliary states count within availability, and availability is a direct product of idle time,[1] so work done in that state is idle by definition. Occupancy states the consequence for one metric; this page states it for the ledger and the requirement.
The contamination problem
Three conditions produce the bleed together, and removing any one removes it: the agent is in a state the telephony platform counts as available or auxiliary; the agent works an item on a tool that emits no state the ledger reads; and no routing rule reserves the agent's time for the asynchronous work, so the item is worked in the gaps between calls. The condition is common wherever a backlog has been placed with a voice team to use the idle time (Deferrable Work and the Idle Cushion). The bleed runs one way, from the measured channel's idle bucket into the unmeasured channel's production, and the voice platform cannot show it, since it records exactly what it was built to record.
What it does to the numbers
Availability is overstated by the handling time of every asynchronous item worked inside it, and the real-time function reads the excess as capacity it can route calls into.
Occupancy (OCC-01 on the ledger: in-work time ÷ productive time) is understated, because the email minutes sit in productive time in the denominator while their in-work time is missing from the numerator.
The voice requirement. An interval model such as Erlang C computes the agents that must be available, and the schedule supplies that many. If a share of their time goes to email, effective voice supply is smaller than the schedule says, the service level is missed when email is heaviest, and the reforecast raises the voice requirement to compensate: the wrong remedy for the right symptom.
The asynchronous requirement. The model on Workload Model for Asynchronous Channels needs handling minutes per item. Where the minutes are in nobody's bucket the requirement rests on an estimate, [E] at best, and the hour voice already counted as supply is counted again as email capacity: the double count Multichannel Capacity Model Assembly warns of.
The detection tests
Three tests find the bleed from data most functions already hold; GP-605 runs them on every blended cohort. Test 1 is T-06 of the Reconciliation Tests for Time Accounting, the availability-bleed test on available-idle time; a failure of T-05, item count against time, in which implied handling time exceeds recorded in-work time, points here; tests 2 and 3 are this page's own.
| Test | Computation | A positive | Needs |
|---|---|---|---|
Items closed per available-idle hour (T-06) |
items closed by the cohort per day ÷ its available-idle hours that day (productive time minus voice in-work time) | a cohort idle in the state closes nothing; a steady rate is production inside it | item counts with closing timestamps; state time by agent and day |
| Focus time against state time | the time an item had focus, by agent and day, against the same agent's available-idle time | focus time inside available-state windows; the overlap is the bleed, measured | a desktop or work-item instrument, with the statutory flag on GP-605 observed |
| A residual that moves with volume | the idle residual (available-idle time minus voice handle time) by day, regressed on items closed by day | a residual that rises on the days more items close; an uncontaminated residual moves with voice volume, not email volume | daily residual and item counts; a few weeks of history |
The first test runs first; the second is the only one that yields a per-item handling time; the third is an inference graded [E] with the regression's range, valid only under the conditions Email Handle Time Measurement places on statistical inference.
The separation methods
Separation puts the asynchronous minute in its own bucket. Three methods do it, in rising order of cost and precision.
Dedicated work states. A telephony auxiliary code reserved for asynchronous work, carrying the attributes Activity-Code Consistency requires (not available, productive, in-work for the named channel), entered when the item is opened. It is the agent-toggled timer, with its biases: forgetting, and a pace that responds to incentives and to how much work is waiting.[2] It fixes the ledger without fixing the measurement.
Work-item timers. A timer in the case or email tool that opens on first touch and closes on completion or pend, reconciled nightly against the telephony state so that a minute carrying both a call and an open item is attributed by a stated rule: the transaction-level capture Email Handle Time Measurement recommends, and the only method that yields a measured [M] handling time per item.
Reserved time blocks through routing. A rule that takes the agent out of the voice pool for a scheduled block, or a threshold rule that releases agents to asynchronous work only when free voice agents exceed a reserve. The reserve is the control variable. The call-blending model of Bhulai and Koole assigns agents to a lower-priority class, outbound calls in their case, only when free agents exceed a threshold,[3] and Legros, Jouini and Koole apply an adaptive threshold of the same form to calls and emails, reserving agents for calls under time-varying arrivals while maximising email throughput within a call-wait limit.[4] A share of the available state is not a policy; a reserve rule is.
Interim reporting until fixed
Between the audit and the fix, the function reports the blended cohort's occupancy both ways and grades each: as measured, bleed left in, [C]; and corrected, the suspected bleed added to the numerator, [E] with the range the detection test supplies. The difference is published as the bleed residual, in hours, beside the daily figure. The Source of Truth Register row for the cohort's email handling time (ST-007, against AHT-02, RG-008) carries the test as its known gap, and Data Quality Gates for Planning Inputs admit the corrected figure only with its range. The two figures are never averaged; a precise number that measures the wrong thing is the trap the framework exists to close.
Worked example
A blended cohort of 40 agents [E] works voice and email on one telephony platform and a case tool with no timer, in the week beginning Mon 5 Oct 2026. The ledger records 1,600 hours of productive time (tier 5) [E] and 880 hours of voice in-work time [E]: reported occupancy (OCC-01) is 880 ÷ 1,600, or 55 percent [C], and the apparent available-idle time is 720 hours [C]. The case tool records 3,600 emails closed [E], nearly all inside available-state windows.
Test one (T-06): 3,600 items ÷ 720 available-idle hours = 5.0 items per hour [C], on a cohort that should close none while idle. Test two, run with a focus-time instrument on eight agents under agreed terms, gives 9.5 minutes per item [E], range 8 to 11, so implied email in-work time is 3,600 × 9.5 ÷ 60, or 570 hours [C], range 480 to 660. Test three: the daily residual rises with closures [E].
Corrected occupancy is (880 + 570) ÷ 1,600, or 91 percent [E], range 85 to 96, against the reported 55 percent [C]; the true available-idle time is about 150 hours [E], not 720. Effective voice supply was 40 × (1,030 ÷ 1,600), about 26 agent-equivalents [E] against a schedule of 40, which is consistent with the missed service level on the two heaviest email days [E]. The function adopts a dedicated work state the following Monday and proposes a work-item timer through GP-606. Every number above is synthetic.
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 function that runs the first test on its blended cohorts and reports occupancy both ways is at Level 2; one with a separation method, a measured handling time per item and a reconciled ledger is at Level 3, because a reserve rule can only be released on a ledger that shows what it reserved; one that carries the bleed residual as an input distribution rather than a point correction is at Level 4.
See Also
- Workforce Data Definitions Framework — the hub: the ledger, the registers and the channel pages this practice serves
- Agent Time Accounting Model — the ledger in which every minute lands in exactly one bucket
- Occupancy — the metric the bleed understates
- Deferrable Work and the Idle Cushion — placing deferrable work on idle voice capacity, from the placement side
- Reconciliation Tests for Time Accounting — the availability-bleed test among the others
- Time and Production Accounting Audit (generic process) — GP-605, which runs the tests on every blended cohort
References
- ↑ LaReau, T. (2012, updated 2018). "Constructive use of occupancy". ICMI. icmi.com.
- ↑ Hasija, S., Pinker, E., & Shumsky, R. A. (2010). "Work expands to fill the time available: capacity estimation and staffing under Parkinson's Law". Manufacturing & Service Operations Management 12(1), 1–18. doi:10.1287/msom.1080.0250.
- ↑ Bhulai, S., & Koole, G. (2003). "A queueing model for call blending in call centers". IEEE Transactions on Automatic Control 48(8), 1434–1438. doi:10.1109/TAC.2003.815038.
- ↑ Legros, B., Jouini, O., & Koole, G. (2015). "Adaptive threshold policies for multi-channel call centers". IIE Transactions 47(4), 414–430. doi:10.1080/0740817X.2014.928965.
