Workload Units and the Roll-Up to FTE

From WFM Labs

Workload Units and the Roll-Up to FTE is the ladder of units by which a workforce function converts a demand forecast into a count of full-time equivalents: contacts, to work items, to handling minutes, to productive hours, to scheduled hours, to paid hours, to full-time equivalents. Each rung is a conversion with a factor, and the page's rule is that every factor carries a grade and cites the register row it came from, so that the resulting FTE figure can be walked back down the ladder to the measured time it rests on. It is part of the Workforce Data Definitions Framework. The ladder is the Agent Time Accounting Model read upward: in-work time to productive time to logged, scheduled and paid time are the ledger's tiers, and the factors between rungs are the ledger's gaps expressed as ratios. Shrinkage, Occupancy and Average Handle Time are defined on their own pages; here they appear as factors with a provenance.

The roll-up is the oldest calculation in workforce planning, taught as base staff grossed up by a rostered-staff factor,[1][2] and the queueing literature supplies the one rung that is not a multiplication: for a synchronous channel, handling minutes become required hours through a staffing model whose output embeds an occupancy the planner does not choose.[3] The framework adds the data-management discipline that lineage is recorded per transformation, not per report.[4]

The unit ladder

Rung One unit is Conversion to the next rung Factor Typical grade Register row cited Ledger tier
Contacts one customer interaction as CON-01 defines it for the channel contacts × items per contact items per contact: 1.0 for voice by definition; above 1.0 where an email thread or a case spawns several items; below 1.0 where several contacts resolve to one item [M] where the work-item tool links items to contacts; [E] otherwise CON-01; the source-of-truth row for the contact count none; demand side
Items one unit of production: a contact handled, a session closed, an item fulfilled items × handling time per item handle time per item by channel; for chat, focus time per session; never elapsed time [M] on voice; [M] or [E] on asynchronous channels by instrument (Email Handle Time Measurement) AHT-01 (session channels) / AHT-02 (asynchronous) tier 6 per unit
Handling minutes minutes of in-work time ÷ 60, then to required productive hours: through the queueing or simulation model for synchronous channels; ÷ the target in-work share of productive time for asynchronous channels the model's implied occupancy, or the target in-work share of productive time [C] from a model whose inputs are graded; the target is [A] OCC-01; the model's lineage card tier 6 to tier 5
Productive hours hours in productive states, in-work or available ÷ (1 − in-schedule shrinkage) the share of scheduled hours lost to non-productive states and to scheduled-not-logged time [M] from the ledger's tiers 2 to 5 SHR-01, in-schedule component tier 5 to tier 2
Scheduled hours hours on the published roster ÷ (1 − off-schedule shrinkage) the share of paid hours not rostered: leave, paid absence, holidays [M] from payroll against the schedule SHR-01, off-schedule component tier 2 to tier 1
Paid hours hours paid in the period ÷ paid hours per FTE per period a contractual constant by country and contract type; SET LOCALLY [M] from the contract of record; varies by jurisdiction the function's FTE definition row tier 1
FTE one full-time paid position for the period — — the weakest grade on the ladder — —

Two rungs deserve a note. Splitting shrinkage into in-schedule and off-schedule components follows the ledger; a single factor is their product and is legitimate when SHR-01's denominator is paid time, but it hides which tier a change came from. And paid hours per FTE is the rung most often left implicit: a function spanning jurisdictions with different standard weeks has as many FTE definitions as jurisdictions, and a consolidated figure states the mix. FTE measures paid hours and headcount counts people; the two reconcile only through the perimeter of Headcount Reconciliation in Service Operations.

The roll-up table

One asynchronous channel, one week; numbers synthetic [E], computed values [C] on [E] inputs.

Rung Value Grade Factor applied Source row
Contacts forecast 10,000 [E] items per contact 1.20 [E] CON-01; forecast lineage card
Items 12,000 [C] handle time per item 9.0 min [E] AHT-02 (RG-008); work-item timer row
Handling minutes 108,000 [C] ÷ 60 (exact)
Handling hours 1,800.0 [C] ÷ 0.85 target in-work share of productive time [A]; plausible range 0.80–0.90 OCC-01 (occupancy form); asynchronous model lineage card
Required productive hours 2,117.6 [C] ÷ (1 − 0.22) in-schedule shrinkage [E] SHR-01 in-schedule component
Scheduled hours 2,714.9 [C] ÷ (1 − 0.10) off-schedule shrinkage [E] SHR-01 off-schedule component
Paid hours 3,016.6 [C] ÷ 37.5 paid hours per FTE per week [M], SET LOCALLY from the contract of record FTE definition row
FTE required ≈80 FTE (80.4 [C]) weakest input [A] the 0.80–0.90 range on the target gives about 76–85 FTE [C]

For a synchronous channel the fourth row is the staffing model: contacts and handle time per interval enter an Erlang C, Erlang-A or simulation model with the service goal, which returns required productive hours per interval with occupancy as an output. Service Level Mathematics for Voice, Chat and Email carries the models; Multichannel Capacity Model Assembly carries how channel requirements combine before the shrinkage rungs are applied to the combined figure, and where applying them per channel double counts the shared shrinkage of a blended pool.

Grade propagation

The FTE figure inherits the weakest grade on its ladder, the rule Human Gates and Number Grades states for any derived number and Data Lineage Card for Planning Models records per model. In the example the weakest input is the asserted in-work share target, so the result is published as ≈80 FTE (80.4 [C]), weakest input [A], with the 76–85 FTE range the target's plausible band implies: a reader who sees a one-decimal figure assumes a measured chain, and the ladder shows it is not. A factor re-measured since the last cycle raises the ladder's grade; a factor carried across a platform or channel change drops to [A] under the register's carry rule until re-measured.

The audit trail

A roll-up is published with its ladder, never as a bare FTE count. The trail has four parts.

  1. The ladder table itself, with a value, a grade, a factor and a source row per rung, and the version of each register entry cited.
  2. The reverse walk. For the period just closed, the ladder is walked downward from actual paid hours (ledger tier 1) through actual scheduled, productive and in-work hours to actual items and contacts. Each actual is compared to the rung the plan predicted. A rung that will not walk back, where the plan's factor and the ledger's actual differ by more than the local tolerance, is a finding against that factor's source row, and the item-count-against-time test on Reconciliation Tests for Time Accounting is the formal version of the check.
  3. The change record. A factor changes between cycles with a reason and a date, so that two months' FTE figures decompose into volume, handle-time, shrinkage and definitional change; factors that drift silently report every definitional change as an operational one.
  4. The model inventory entry. Each model that converts handling minutes to required hours is listed with its lineage card, so the FTE figure's dependence on a queueing assumption or simulation seed is on the record.

The trail is what makes the roll-up a planning instrument: a stakeholder who disputes the FTE count is shown the rung and the row, and the dispute becomes a register change through Definitions and Source-of-Truth Change Control (generic process) rather than a negotiation over a total.

Where the ladder breaks

  • Contacts counted as items. A contact forecast applied to a per-item handle time misstates workload by the items-per-contact factor; the CON-01 question is settled before the first rung.
  • Elapsed time as handling time. An item open for two days consumed perhaps twenty minutes of agent time; the rung reads touch time, and an open-to-close duration in its place inflates the requirement by the pend time.
  • Concurrency ignored. A chat session's elapsed time is not an agent's focus time; Chat Concurrency and Handle Time gives the correction.
  • A blended handle time across channels. One average over voice, chat and email has undefined units; the ladder runs per channel to the required-hours rung and combines there.
  • An occupancy target as a factor for a synchronous channel. Occupancy is an output of the staffing model at a given service goal and pool size; dividing by a target skips the model and misstates small pools in particular, as Occupancy and Pooling Architecture in Service Workforces explain.

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 Level 2 function runs the ladder with a single shrinkage factor and a measured voice handle time, and publishes the FTE total. Level 3 runs it per channel with the two shrinkage components from the ledger, publishes the ladder with grades, and performs the reverse walk each cycle; the automation layer depends on that ladder, because a rule that adjusts staffing intraday must read the same factors the monthly plan was built on. At Level 4 the ladder is a lineage object the capacity model emits automatically, and a change to any register row re-runs it.

See Also

References

  1. ↑ Reynolds, P. (2003). Call Center Staffing: The Complete, Practical Guide to Workforce Management. The Call Center School Press (the chapters on workload, base staff and the staffing roll-up).
  2. ↑ Cleveland, B. (2012). Call Center Management on Fast Forward: Succeeding in the New Era of Customer Relationships, 3rd ed. ICMI Press, part 3.
  3. ↑ Gans, N., Koole, G., & Mandelbaum, A. (2003). "Telephone Call Centers: Tutorial, Review, and Research Prospects". Manufacturing & Service Operations Management 5(2), 79–141 (the Erlang-C staffing model and square-root safety staffing).
  4. ↑ DAMA International (2017). DAMA-DMBOK: Data Management Body of Knowledge, 2nd ed. Technics Publications, chapter 12 (metadata management: lineage) and chapter 13 (data quality). dama.org.
Retrieved from "https://wiki.wfmlabs.org/w/index.php?title=Workload_Units_and_the_Roll-Up_to_FTE&oldid=5477"