Wiki:Packs/Business Travel Seasonality and Supply Flex

Pack
ID	CP-WFM-026
Name	Business Travel Seasonality and Supply Flex
Slug	business-travel-seasonality-and-supply-flex
Type	Pack
Domain	WFM
Status	Active
Version	1.0
Contents	1 instruction block + 4 reference files + briefing prompt
Builds on	Business Travel Demand Seasonality Seasonal Capacity Contracting with Outsourcing Partners Working-Time Averaging and Annualized Hours in Contact Centers Supply Elasticity in Workforce Planning Outsourcing as a Risk Lever The Vendor as Relief Valve

A pack is a deployable set for a Claude project: one instruction block pasted into the project's custom instructions, plus reference files saved as Markdown and uploaded as project knowledge. See Wiki:Packs.

This pack supports a workforce leader who buys outsourced contact-center capacity for a business-travel operation and wants that capacity to move with the week-by-week demand curve instead of sitting flat through it. It does three jobs: it builds and maintains a macro weekly seasonality curve from public sources and from the operation's own segment books; it evaluates partner contract terms for the flexibility they actually give, and designs the terms to ask for; and it turns location rules and market rates into a view of where a swing can be locked up or down, for how long, and at what price.

The problem it addresses

Outsourced capacity is bought in fixed blocks. Billing typically moves one or two percent month to month while business-travel demand moves by ten to twenty, with a December trough deep enough that no hours model fixes it. The only ways to change a block today are to add one and keep paying for it, or to cut one and spend eight to sixteen weeks of training when the work returns. Meanwhile the demand shape is largely predictable: roughly nine-tenths of week-to-week variance is seasonal, the rest is disruption.

The pack treats flexibility as something that is bought, measured and placed. It is bought as a band of productive hours around a median base, priced in the contract. It is measured against one definition of a productive hour. It is placed in the countries whose working-time law lets hours flex without a premium, and the premium is paid knowingly where it does not.

What this pack does not restate. Why supply elasticity sets the ceiling on forecast value is covered at Supply Elasticity in Workforce Planning; when outsourcing is a risk lever rather than arbitrage at Outsourcing as a Risk Lever; the valve design rules at The Vendor as Relief Valve. The pack's files are derived from the three articles in the infobox and summarize the rest in a line where a block needs them.

When to use it

• Preparing a flexibility proposal or a renewal negotiation with an outsourcing partner
• Building or refreshing the seasonal demand curve for a travel operation, before or after internal data are available
• Deciding which sites and countries should carry the seasonal swing, and which should hold the base
• Reading a partner's market pricing study and turning it into comparable terms
• Sizing the December plan, the September peak and the disruption reserve as separate problems

Not needed for ordinary capacity planning where demand is stable or where capacity is entirely in-house with no contractual flex to design.

How to deploy

1. Create a project, named for the operation rather than the method
2. Copy Block 1 into the project's custom instructions
3. Save Blocks 2–5 under the filenames given in their headings and upload as project knowledge, together with any working paper, footprint, billing extract or partner pricing study the operation holds
4. Start a conversation with the briefing prompt in Block 6

Block 1 — Project instructions

This project designs supply-side flexibility for a business-travel contact
operation that buys part of its capacity from outsourcing partners. Demand
moves through a repeating weekly curve; supply is bought in fixed blocks.
The work is to measure the curve, decide which levers carry which part of
the year, and turn partner contracts and site choices into a priced band of
productive hours that moves with the curve.

Governing idea: flexibility is bought, measured and placed. Bought as a band
of weekly productive hours around a median base, priced in the contract.
Measured against one definition of a productive hour. Placed where
working-time law lets hours flex without a premium, with the premium paid
knowingly where it does not.

WHICH FILE TO OPEN
  Build or refresh the weekly curve; blend books;
  translate bookings into contacts; express it so
  it survives internal data                        -> seasonality-evaluation.md

  Read a contract for the flex it gives; design the
  terms; price a band; overtime vs hire vs rehire  -> partner-flex-terms.md

  Where the swing should sit; read a pricing study;
  country hours-law verdicts; lock windows by site -> location-rules-and-rates.md

  The levers, what each carries, how they combine
  across the year, and the ranking                 -> lever-inventory.md

Open one. Answer from here if none is needed.

DISCIPLINES
- Index, not heads; any headcount is an illustration on a stated, assumed base.
- Anchor on what is measured (quarterly filings); label weekly shape, elasticity and disruption share as inferred until internal data replace them.
- Every output carries a basis table: measured / inferred / assumed, and what moves when internal data arrive.
- Never shape a business-travel curve on leisure-dominated series such as airport throughput.
- Comparability before allocation: no partner earns volume for flexibility until quality is measured like for like.
- Rate is a column of the site scorecard, never the sort key; buy the hour that meets the standard in the window it is needed.
- Release-and-rehire is a last resort: it pays only for troughs longer than the replacement cost (about 14-22 weeks); most seasonal troughs are shorter.
- Judge a shared bench on a bad day: disruption hits every travel client of a shared partner at once.
- Clear the swing across the operation's own books before flexing the partner estate.
- Rates, partner names and per-vendor spend stay in the project's files; circulated outputs carry indices and models.

OUTPUT
Lead with the answer. Show the arithmetic for any sizing, band or cost
figure. Give low, expected and high cases with the load-bearing assumption
named. Quote the clause or section where a contract or a law is the source.
Where the data cannot settle a question, say what would settle it.

Source: Wiki:Packs/Business Travel Seasonality and Supply Flex (CP-WFM-026) v1.0

Block 2 — seasonality-evaluation.md

# Evaluating business-travel seasonality

*Figures marked [measured] come from a public filing or statistical series.
Figures marked [estimated] are judgment or arithmetic on stated assumptions.
Nothing here replaces an operation's own weekly contact history; the method
exists so that the public-data edition and the enterprise-data edition read
the same way.*

## What the curve is for

The seasonal curve is the demand side of a supply-flex design. It answers:
how far above and below the average week does required capacity move, for
how many weeks at a time, and how much of that movement is predictable?
Those three answers size the band, the December plan and the event reserve.

## Sources, by grain and bias

| Source | Gives | Bias to correct for |
|---|---|---|
| Quarterly transaction value in the filings of listed travel management companies | The only published quarterly anchor; in one application, roughly Q1 106, Q2 103, Q3 101, Q4 90 on an average quarter of 100 [estimated from public filings, 2023-24 average]; rebuild from the filings relevant to the market | Price and mix; within-year growth lifts later quarters; an acquisition inside the series breaks it |
| Quarterly revenue, same companies | A cross-check | Fees and lags damp it to about 103-97; understates volume |
| Agency settlement releases (monthly trips) | December roughly 10% below November; January roughly 40% above December | Includes leisure and online agencies |
| Corporate trip indices | December 20-21% below November in one year | Single vendor; irregular |
| European household tourism statistics | Business trips in July-August at about two-thirds of the monthly average | Survey artifacts in some months |
| Business hotel booking indices | Europe August far below June; September the busiest month | Hotel only |
| Airport security throughput | Daily, clean, free | WRONG POPULATION: leisure-dominated, no December trough. Use for day-of-week travel shape only, never for the business curve |
| The operation's own weekly contacts and bookings by segment book | The truth | Replace every inferred figure above with it when available |

## Method

1. Anchor the quarters on the transaction index, averaged over two or more
   years with no structural break (an acquisition in the series breaks it).
2. Set a provisional weekly booking index from the monthly and event
   evidence: a steep two-week climb in January; United States holiday weeks
   (Memorial Day, 4 July, Labor Day, Thanksgiving); the European summer dip
   in weeks 27-34; the September peak; the slide from week 49.
3. Rescale each quarter so its 13-week mean equals the anchor; normalize the
   year so the average week is 100.
4. Build a travel index by passing the booking index through a lead-time
   kernel (0-6 weeks, mean about 3 weeks). Bookings lead departures.
5. Build the contact index as the mean of booking and travel indices:
   new-booking contacts follow bookings; servicing, changes, cancellations
   and disruption follow travel.
6. Grade each segment of the year: level confidence and shape confidence
   separately. The last two weeks of the year and the Thanksgiving week are
   the least supported and matter most to the December plan.

## Blending segment books

Operations usually serve several books: brands, client segments, channels
or regions, each with its own curve. Their peaks do not fully coincide. Build the macro curve as a
weighted blend and keep the components:

| Week | Book 1 index | Book 2 index | Book 3 index | Weight 1 | Weight 2 | Weight 3 | Blended |
|---|---|---|---|---|---|---|---|

Weights are each book's share of contact workload (not revenue). The gap
between the blended curve's swing and the sum of the books' swings is the
netting dividend: the part of the seasonality the operation can clear
across its own books before any partner lever is needed.

## Regional overlay

The global anchor is a blend; applying a regional split does not change the
quarterly means, but it changes site-level staffing.
- North America: mild July-August dip; September-October peak; Thanksgiving
  and 20 December-1 January troughs.
- Europe: July-August about a third below average; September at the year's
  high; the booking dip starts in late June because bookings lead travel.
- Asia-Pacific: Lunar New Year, Golden Week, Obon; December is not a trough
  in most markets. Thin data; calendar knowledge only.

## From bookings to contact workload

Public data cannot give contacts per booking. Model it as an elasticity e,
the share of workload that moves with the booking curve: low 0.5, expected
0.75, high 1.0 [estimated]. Required capacity index for week w:

    R(w) = 100 + k x (C(w) - 100),  k = amplitude x e

with amplitude 0.6 / 1.0 / 1.25 for low / expected / high. Worked example,
expected case, week 52 at C = 64: R = 100 + 0.75 x (64 - 100) = 73, that is
27% below the average week. Week 39 at C = 115: R = 111.

Replace e with a regression of weekly contacts on weekly bookings as soon
as internal data exist. Expect e below 1 where back office, account
management and minimum staffing do not move with bookings.

## Disruption and the predictable share

Treat disruption as a separate stream. If it is about a tenth of annual
contacts, a moderate regional event (x2.5 for three days) adds about 6% to
that week; a severe one (x4 for four days) about 17% [estimated]. With a
seasonal weekly standard deviation near 8% and an event standard deviation
near 2.5%, roughly nine-tenths of week-to-week variance is predictable.
Staff to the seasonal curve; hold an event reserve of 6-10% of workload
reachable within a day; plan weeks 48-52 as a separate workload plan.

## Reading a partner's market study alongside the curve

A partner's pricing or seasonality study arrives in its own units. Convert
it before comparing: express its volumes as an index on the average week,
its rates as an index on the operation's reference rate for the same
country, tier and shift, and its flex terms in the vocabulary of
partner-flex-terms.md (commitment unit, notice up/down, corridor, charge).
Keep the study's own figures in a source table with the study named; never
merge them silently into the curve.

## Output conventions

Every deliverable carries:
- a basis table: measured / inferred / assumed, with what moves when
  enterprise data arrive and what stands;
- the curve as an index with low / expected / high cases;
- any headcount as an illustration on a stated, assumed base, with one line
  showing the same percentages on a second base;
- the gap against a flat supply line, which is the argument, rather than a
  required-capacity figure, which nobody staffs to;
- an edition label (public-data edition; enterprise-data edition) and the
  list of what changes between them.

## Checklist before the curve is used

- Anchor years have no acquisition or structural break.
- Airport throughput was not used to shape the business curve.
- Each segment of the year carries a level and a shape confidence.
- The elasticity and the disruption share are labeled [estimated].
- The December weeks are flagged as the least supported and most material.
- Segment-book components are kept, not only the blend.

Block 3 — partner-flex-terms.md

# Partner contract flexibility: reading it and designing it

*Design patterns from practice, not published standards. Economics are
arithmetic on stated assumptions and are marked [estimated].*

## Step 1: score what the contract gives today

Complete one row per partner, per contract, from the clauses, not from
memory. Cite the clause.

| Field | What to record |
|---|---|
| Commitment unit | FTE block / FTE per month / productive hours per week / transaction / outcome |
| Productive-hour definition | What counts (logged-in, handle, available), who reconciles, from which data |
| Minimum term | Months |
| Notice to go up | Weeks, and any cap on the size of a step |
| Notice to go down | Weeks, cost, and any floor; this is the half partners resist |
| Corridor or deadband | Baseline, tolerance, what happens outside it |
| Additional / reduced resource charges | Tiers, prices, credits |
| Overflow or bench terms | Rate, certification, ramp time, data access |
| Shift and tier premiums | After-hours blocks, proficiency tiers, as rate multipliers |
| Flexibility in force | Which of models A-F below are live; usually none except D on small contracts |

Compare the scorecard with billing: if billed FTE moved less than two
percent across a season in which demand moved ten or more, the flexibility
on paper is not being used, or does not exist.

## Step 2: the commitment unit decides what can flex

Moving the unit from FTE to productive hours per week inside a band is the
one change that lets the other levers work. It needs one productive-hour
definition both sides read from the same data. Under FTE-per-month pricing,
ask whether flex hours are included at no charge: the partner's marginal
cost of an extra scheduled hour is mostly shrinkage.

## Step 3: the lock ladder

| Horizon | Locks | Tolerance |
|---|---|---|
| 90 days | Outlook by pool | Indicative |
| 45 days | Weekly band | +/-10% |
| 21 days | Daily hours | +/-3% per day |
| 7 days | Daily schedule | Fixed |
| 1 day | Half-hourly staffing | +/-3% per half hour |

Add an over-lock billing cap (for example 5%) and a protected ramp-down.
The 45-day lock is the band; the daily and half-hourly locks are how it is
executed. The buyer's side is a rolling 13-week forecast by pool, refreshed
weekly. A partner will not carry a band it cannot plan against.

## Step 4: name and meter two layers

- Committed flex: the band the partner staffs to; priced into the base rate
  with a small premium above base and a small discount for released hours,
  so the partner is paid for carrying the option.
- Contingent flex: bench, surge hours, overflow above the band; a standing
  option fee, exercised with notice.

Ask for the standard mechanism by name: additional and reduced resource
charges with a deadband (baseline, tolerance of +/-5-10%, tiered declining
charges above, credits below). The reduced resource charge is the test of
whether the band works in both directions.

## Step 5: choose the model per pool

| Model | Mechanism | Fits | Watch for |
|---|---|---|---|
| A. Flex corridor on FTE | Base +/- x% at 30/60/90 days | Any partner; the minimum ask | Corridors that only go up |
| B. Annualized hours, surge band | Hours per head move inside an annual total (e.g. 32-46 on a 38 base) | Pools in averaging-law countries | Country law; consent; troughs are pay cuts unless pay is annualized |
| C. Bench or burst hours | Partner's cross-client bench at a bench rate | Multi-client sites | Certification; data access; correlated surge |
| D. Per transaction | Price moves with volume | Back office, email, simple voice | Peak priced into the unit; unit definition |
| E. Shaped volume commitment | Buyer commits annual hours; partner shapes to the curve | Large deferrable blocks | Partner charges for hiring risk; needs a trusted forecast |
| F. Growth for flex | Volume to the partner that carries the swing | Deepest bench in the main language | Concentration; comparability before allocation |

Precondition for all six: price the few genuinely dedicated seats and pool
the rest. Seats are usually designated on paper and shared in operation.

## Step 6: the median-base, surge-hours proposal

Hire and train at the median week. Peak weeks rise toward 45 (50 for a
one-to-two-week burst) where law and the person allow; trough weeks fall
toward 30-32. Pay follows hours inside an annual total: no overtime
trigger, no pay cut. The partner commits productive hours per week inside
the band, agreed a quarter ahead from the forecast.

Needs from the partner: a band per pool in weekly productive hours; a price
for the band, not only the hour; cross-skilling as a deliverable (two
skills per agent in a closed chain gives most of the flexibility of
universal agents); committed and contingent layers metered; one
productive-hour definition.

What the buyer brings: the 13-week forecast; the definition and the data
annex to measure it; a certified record of who is eligible for which
queue; growth where it is earned; a willingness to pay for the option.

## Step 7: economics

Cost per productive hour, base = 1.00 [estimated]:

| Peak source | Cost |
|---|---|
| Hours-flex inside an averaging scheme | 1.00 |
| Overtime at +25% | 1.25 |
| Overtime at +50% | 1.50 |
| Overtime at +100% | 2.00 |
| Seasonal hire, 6 training weeks, 12 peak weeks | 1.76 |
| Seasonal hire, 8 and 12 | 1.96 |
| Seasonal hire, 8 and 20 | 1.65 |

Seasonal hire formula: (T + P) / (0.85 P), with 85% of tenured
productivity during the peak; recruiting cost excluded, which flatters the
hire. Crossover peak length: P = T / (premium x 0.85 - 1). At a 1.5
premium the hire breaks even only for peaks longer than about 22 weeks
(T = 6) to 29 weeks (T = 8), so hours-flex at ordinary overtime rates wins
for any seasonal peak. At double time the crossover is about 9-11 weeks,
so where the premium is 2x a head-add is competitive for peaks longer than
about twelve weeks.

Troughs: replacement = training weeks (8-16 for skilled travel work) +
recruiting (about 2 weeks' pay) + ramp shortfall (about 4 weeks) = roughly
14-22 weeks of loaded pay. Retaining through a trough costs at most the
idle fraction of paid hours, less if hours can fall to 30. Release pays
only for troughs longer than that; the December trough is five weeks.

## Step 8: four questions for the partner

1. What flexibility does the contract give today: unit, notice up and down,
   minimum term, corridor? Good answer: a number for each and the clause.
2. At multi-client sites, what shared or bench capacity exists, how is it
   skilled, how is it priced? Good answer: a bench rate card and ramp time.
3. Would you run a pool on annualized hours with a band, and what does the
   band cost against today's rate? Good answer: a priced corridor and the
   HR rules behind it.
4. Which queues would you put in a cross-trained group, and what would it
   take to certify an agent across them? Good answer: a skill map, a
   training plan in weeks, a proficiency gate the buyer defines.

## Red flags

- A corridor with notice up but not down.
- A bench offer with no certification path or no data access.
- A productive-hour definition that lives only in the partner's system.
- Flexibility offered as goodwill rather than as a clause.
- A per-transaction price that quietly carries the peak in the unit.

Block 4 — location-rules-and-rates.md

# Location strategy for a seasonal swing: rules, lock windows and rates

*Country rows come from statutes and reputable summaries; several rest on
secondary texts. Not legal advice. Confirm with counsel before a figure
enters a term sheet.*

## The question this file answers

Where should the swing sit? Not "where is the lowest rate" but: in which
sites can hours be locked up or down, for how long, at what premium, with
what notice, in the languages and hours the work needs, with how much
exposure to correlated surge. Rate is one column of the scorecard and never
the sort key. Buy the hour that meets the standard in the window it is
needed.

## How the lock window drives the rate

A partner prices three things into an hour: the hour itself, the risk of
carrying people the buyer may not use, and the cost of changing the number.
The longer the buyer can lock capacity (a committed annual volume shaped to
the curve), the less risk the partner carries and the lower the rate. The
shorter the notice the buyer wants (a 21-day daily lock, a bench reachable
in days), the more the partner charges for the option. The site decision is
therefore a trade between lock length and rate, made per pool:

| Lock length | What the buyer gets | What it costs |
|---|---|---|
| Annual committed hours, shaped | Lowest rate; partner shapes to the curve | Hiring risk priced in; needs a trusted forecast |
| Quarterly band (45-day lock) | Hours move inside a corridor at base rate | Small band premium; consent and law per country |
| Weekly / daily lock | Execution of the band | Tolerance charges above the lock |
| Bench, days' notice | Short peaks at a bench rate | Option fee; certification; correlated surge |
| Overtime | Immediate | Statutory premium: 1.25 to 2.0 |

## Site scorecard

One row per site per partner. Fill from the contract, the country table
and the footprint; score, then read the pattern. Do not sum the columns.

| Column | Record |
|---|---|
| Hours-law verdict | Free / free with averaging / needs agreement / premium / blocked |
| Averaging window | Weeks or months over which hours may average, and the daily cap |
| Premium | Statutory multiplier beyond the threshold |
| Notice rules | Schedule publication notice; consent requirements |
| Hours of operation | 24x7 / business hours / after-hours block |
| Languages | Which pools the site can serve |
| Pool size | Heads in the smallest gated pool at the site (small pools cannot absorb a swing) |
| Bench | Does the partner run a cross-client bench here; certification path |
| Correlated-surge exposure | Share of the site's other clients in travel |
| Rate index | Site rate / reference rate for the same tier and shift (index, not currency, in anything circulated) |
| Band price | Premium for hours above base; discount for released hours |
| Follow-the-sun fit | Which other sites share its language and can cover its shoulders |

## Country reference (hours law; 30-to-50 without a premium?)

| Country | Standard week | Averaging | Premium | Verdict |
|---|---|---|---|---|
| India | 48 | None found; new codes since late 2025 | 2x with consent; quarterly cap (sources conflict) | Premium |
| Philippines | 48 (8/day) | Voluntary compressed week to 12 h/day | +25% over 8/day; +10% night | Premium above 8/day; sixth day or overtime for 50 |
| Morocco | 44 | Annual distribution exists; unverified | +25% day, +50% night; cap 80-100 h/yr | Premium; tight annual cap |
| Romania | 40 | 48 avg over 4 months (6-12 by agreement); in the contract | Time off in 90 days, else +75% | Needs agreement |
| Bulgaria | 40 | Summary calculation 1-4 months (12 by agreement) | >= +50%; caps 6/wk, 30/mo, 150/yr | Needs agreement |
| Sweden | 40 | 4-week averaging only | By collective agreement; 200 h/yr | Partial |
| Canada (Quebec) | 40 | None general | +50% over 40; refusal beyond 2 h | Premium |
| South Africa | 45 | To 4 months by collective agreement | +50%; max 10/wk | Needs agreement |
| UK (Northern Ireland) | None statutory | 48 avg over 17 wks; opt-out | None statutory | Free contractually |
| Ireland | None statutory | 48 avg over 4 months | None; 24 h notice of hours | Free contractually |
| Poland | 40 | Settlement period 4-12 months | +50% (+100% night/holiday); 150/yr | Needs agreement |
| Czech Republic | 40 | Working-time accounts 26-52 wks; 2 wks' schedule notice | >= +25% or time off | Free with averaging |
| Turkey | 45 | Equalization 2-4 months; max 11/day | +50%; cap 270/yr | Free inside the window |
| UAE | 48 | None; 144 h per 3 weeks | +25% (+50% night) | Premium |
| Spain | 40 annual avg | 10% irregular by default; more by agreement | Ordinary rate or time off; 80/yr | Blocked by the 9-h day unless agreed |
| Italy | 40 | 48 avg over 4-12 months; annual by agreement | By collective agreement; 250/yr | Needs agreement |

Sustained 50-hour averages are unlawful under the 48-hour average cap in
the EU and the UK (absent opt-out), under Turkey's 45-hour equalized norm,
and under the UAE's 144 hours per three weeks. Plan 45 as the sustainable
peak for at most four or five consecutive weeks and 50 as a one-to-two-week
burst; one practitioner account reports absence rising after five or six
weeks at 45 [estimated as a rule].

Items for counsel: the Indian overtime cap under the new codes; Morocco's
annual distribution; Czech and Bulgarian premiums; expiring sector
agreements in Spain; whether UK zero-hours rules reach Northern Ireland.

## Placing the swing: a reading pattern

- Premium regimes (India, Philippines, Morocco, Quebec, UAE) hold the base
  and the back office. Their surge is a compressed-week and sixth-day
  design with a known premium, or deferrable work that queues in peaks and
  clears in troughs.
- Averaging regimes (Romania, Bulgaria, Czech Republic, Poland, Turkey,
  Italy, South Africa) and no-premium regimes (Northern Ireland, Ireland)
  carry the hours band, subject to pool size: a 60-person language pool
  cannot absorb a swing that is 40% of itself, whatever the law allows.
- Language gates where the law would allow a band are the binding
  constraint for European languages; follow-the-sun works for English and
  not for them.
- The December trough is placed nowhere: it is a leave and certification
  plan at every site, not an hours plan.

## Reading a partner's market pricing study

Extract into the scorecard, per country, tier and shift: the rate (as an
index on the reference rate); the price of the band (premium above base,
discount below); the price of notice (what shorter lock windows add); any
bench rate and its certification condition; the study's own seasonality
claims as an index on the average week. Keep the study's figures in a
source table that names the study and its date. Where its rates differ
from the contracts in force, record both; the gap is a negotiating fact.

## Rules

- Comparability before allocation.
- Rate is a column, not the sort key.
- Every partner seat is variance absorption or market access; pay for
  elastic low-cost capacity, never a variance premium where a backlog
  absorbs the variance free (fulfillment, email, back office).
- Indices and verdicts travel; rates and partner names stay in the project.

Block 5 — lever-inventory.md

# The supply-side lever inventory

*Reach figures are percentages of the average week's capacity and are
[estimated] from one application of the method. They do not change with
the headcount base; the heads they imply do.*

## Five families

| Family | Levers | What they carry |
|---|---|---|
| A. Hours per head | 38-hour base with a flex band; annualized hours; leave and voluntary time off placed in troughs; overtime as last resort; part-time and variable-hour contracts | Most of the ordinary swing. A 38 base with a 34-42 band carries +/-10.5% inside hours; the same payroll buys about 5% more heads than a fixed 40 |
| B. Who can do what | Clear across books first; the cross-trained group; back office as shock absorber; tiered proficiency; merging small pools on the strength of a certified capability record | Moves the swing from each pool's level to the estate's, where it is small |
| C. Where hours come from | Flex corridor; surge-hours band at the partner; bench hours; per-transaction pricing; follow-the-sun lending; growth for flex | The partner share of the swing and the top of the peak |
| D. Pay and incentive | Opt-in surge pay for peak shifts and disruption days; trough incentives (paid leave, certification time placed in troughs); annualized pay | Makes the hours levers acceptable; turns idle trough hours into certification |
| E. Demand shaping | Automation taking simple contacts first in peaks; a deferrable-work calendar; disruption playbooks | What supply never has to carry |

## Scoring each lever

| Lever | Reach (% of average week) | Speed | Cost of the option | Tenure |
|---|---|---|---|---|
| A1 Hours band 34-42 on 38 | +/-10.5 | Weeks; set a quarter ahead | Design and consent work | Protects |
| A2 Annualized hours | +/-15-20 where averaging is lawful | Weeks | Legal/HR setup; consent | Protects |
| A3 Leave and VTO into troughs | -10 to -30 in trough weeks | Days | None if leave is owed | Protects |
| A4 Overtime | +5-10 for short periods | Days | Premium 25-100% | Neutral |
| A5 Part-time / variable-hour share | +30 on that share | Weeks | Administration | Protects |
| B0 Clear across books | Whatever the books do not share | Monthly | Nothing beyond the process | Protects |
| B1 Cross-trained group | The largest; estate-level | Months to certify, then days | Second-skill training; capability record | Protects and raises tenure value |
| B2 Back office as absorber | +/-10 of front-office capacity where back office is large | Days | Backlog service levels | Protects |
| B3 Tiered proficiency | Raises the usable surge pool | Months | Tier premium already priced | Protects |
| C1 Flex corridor | +/-10-20 of the partner pool | 4-12 weeks | Corridor price | Protects ours |
| C2 Partner surge band | +/-15-20 | Weeks | Band premium | Protects |
| C3 Bench hours | +5-15 for seasonal peaks only | Days to weeks after certification | Bench rate | Protects ours |
| C4 Per transaction | Full, on that work | Immediate | Peak priced into the unit | Partner's |
| C5 Follow-the-sun | +5-10 at the shoulders | Days | Language-limited | Protects |
| D1 Opt-in surge pay | +5-10 on peak days | Hours | Premium only when used | Protects and rewards |

## How they combine across the year

Required-capacity index R, expected case (elasticity 0.75 on the contact
index; average week = 100), from one application of the method.

| Part of the year | R | Levers |
|---|---|---|
| Weeks 1-4, the climb | 75 rising to 97 | Band up; cross-trained group; partner band; surge pay for the first fortnight |
| Weeks 5-22, the plateau | 100-110 | Band at the top; back office lending; bench for the top few percent |
| Weeks 23-26, early summer | 99-103 | Band near base |
| Weeks 27-34, European summer | 93-99 | Leave into the dip at European sites; training time |
| Week 35, the return | about 102 | Band up |
| Weeks 36-44, autumn peak | 99-111 | As spring; corridor if the band is exhausted |
| Weeks 45-47, pre-holiday | 98-102 | Band near base |
| Weeks 48-52, the December hole | 73-94 | Leave promotion; annualized-hours floor; certification time; part-time floor. Hours alone cannot fix it: the hole needs 25-32 hours per head |
| Disruption days | +6 to +17 on the week | Surge pay; bench; playbooks; a 6-10% event reserve reachable within a day |

## Sizing in percentages

Peak: about +11 at the single highest week and +9 across the spring
plateau; a 34-42 band on 38 carries +/-10.5 on its own, leaving the high
case (+19) to the cross-trained group and partner bench. Trough: about -23
in the last two weeks; the 34-42 band's floor carries about -10.5 of the
hours of those on it, and a quarter of the estate taking one week of leave
across the fortnight (-12.5) covers the rest. Event reserve: 6-10, carried by surge pay,
back-office lending and bench, not by the seasonal band.

To illustrate in heads, multiply by an assumed base and say so; show the
same percentages on a second base in the same sentence.

## Ranking logic

0. Clear across books: free; shows how much swing is left.
1. Cross-trained group in the main language, across products: everything
   else gets bigger once it exists. Two skills in a closed chain, not
   universal agents.
2. Hours band with annualized pay: same payroll, more heads, most of the
   swing inside hours.
3. Partner surge band, piloted on one 24x7 pool in the main language
   through a peak: shortest commercial path.
4. The December plan as leave plus certification time.
5. Back office as shock absorber.
6. Opt-in surge pay (needs schedule-offer tooling).
7. Bench hours, after certification and data access.
8. Per transaction, extended from small partners to back office.

## What has to be true first

1. A weekly curve from the operation's own data, by pool, two to three
   years deep.
2. One productive-hour definition across in-house, captive and partner.
3. A capability record of who is certified for what.
4. Country law mapped for every site (location-rules-and-rates.md).
5. The economics checked against the operation's own training length and
   attrition (partner-flex-terms.md, step 7).
6. A rolling 13-week forecast the partners can plan against.

Block 6 — Briefing prompt for the first conversation

Paste this as the first message once the pack files and the operation's own documents are uploaded. Delete any numbered item that was not uploaded.

Alongside the four pack files I have uploaded what this operation holds:
(1) any working paper or prior study on seasonality or supply flex; (2) the
partner footprint, billing extract and rate cards; (3) any partner or
market pricing study; (4) any weekly or monthly series for the books the
operation serves.

Do four things, in order, and stop after each for my review.

1. Seasonality baseline. Using seasonality-evaluation.md, build or restate
   the curve as an index with its basis table. Convert each uploaded
   series to an index on the average week, show them against the baseline
   week by week, and say where they agree, where they disagree, and which
   disagreement matters for the December plan or the autumn peak. Keep the
   components; do not merge them.

2. Contract scorecard. Using partner-flex-terms.md step 1, build one row
   per partner from the documents. Mark every field you could not fill.
   Compare the scorecard with billing and say whether the flexibility on
   paper is being used.

3. Site scorecard. Using location-rules-and-rates.md, build the site
   scorecard with rates as an index on a reference rate I will name. Read
   the pattern: which sites hold the base, which carry the band, which are
   gated by pool size or language, where correlated-surge exposure sits.

4. Proposal shape. Using lever-inventory.md and step 6 of the contract
   file, draft a one-page proposal for a first partner pilot: the pool, the
   band, the lock ladder, the price structure, what the buyer brings, and
   the four questions. Percentages throughout; heads only as a marked
   illustration on an assumed base.

Conventions: American English. Index, not heads. Basis table in every
output. Rates and partner names stay inside this project.

Usage notes

• Sizing. Block 1 is roughly 460 words and loads with every message in the project. Blocks 2–5 total roughly 4,900 words and are retrieved only when the routing table sends the model to them.
• Scope. Method and reference only. No organization's data, rates or partner names; the pack is deployable in any business-travel operation that buys outsourced capacity, and most of it in any seasonal service operation.
• Provenance of figures. Quarterly indices are marked [measured] where they come from public filings or statistical series. Weekly shapes, elasticities, lever reach and the cost comparisons are [estimated]: judgment or arithmetic on stated assumptions from one application of the method. None of the weekly figures have been tested against an operation's own contact history. The pack is built so that they can be replaced without rewriting anything else.
• Legal content. The country tables are a planner's reference, not legal advice; the articles behind them grade each row and list the items for counsel.
• Drift. Blocks 2, 3 and the country table of Block 4 derive from the three articles in the infobox; when an article changes, the matching block is regenerated and the version here is incremented. The site scorecard and lock-length table in Block 4 and the lever inventory in Block 5 consolidate material from Supply Elasticity in Workforce Planning and from one application of the method and have no single article home yet; a lever-inventory article is queued, and until it exists those blocks are maintained here.
• Relationship to existing packs. Wiki:Packs/Planning Under Demand Volatility (CP-WFM-001) covers the residual-variance and measurement-window problem; this pack covers the seasonal, forecastable share and the supply levers that carry it. Wiki:Packs/Sourcing Architecture for an Agentic Service Chain (CP-WFM-007) covers node decomposition and sourcing design; this pack assumes the nodes exist and prices their flexibility.

Change history

Version	Date	Change
1.0	2026-10-31	Initial publication. Instruction block, four reference files, briefing prompt.

See also

• Wiki:Packs/Planning Under Demand Volatility (CP-WFM-001) — structural variance, measurement windows, buffer versus reserve
• Wiki:Packs/Sourcing Architecture for an Agentic Service Chain (CP-WFM-007) — node decomposition and sourcing architecture
• Wiki:Packs/Demand Variance Decomposition (CP-WFM-006) — separating seasonal from residual variance
• Business Travel Demand Seasonality — the demand shape and the method behind it
• Seasonal Capacity Contracting with Outsourcing Partners — commitment units, the lock ladder, pricing models, economics
• Working-Time Averaging and Annualized Hours in Contact Centers — the three legal regimes and the country table
• Supply Elasticity in Workforce Planning — why elasticity sets the ceiling on forecast value
• Outsourcing as a Risk Lever — variance absorption versus market access
• The Vendor as Relief Valve — the valve design rules
• Wiki:Packs — the pack registry