Placement Rules and the Tenure Contract

From WFM Labs


Placement rules and the tenure contract is an operating method for deciding where service work should be performed. It rests on a single discriminating question — can the resolution path be written down before the interaction starts — and on the observation that expertise cannot be purchased through a contract while the rate at which people leave can be.

The method is deliberately narrow. It does not replace a sourcing framework; it supplies the decision instrument that a framework leaves unspecified. For the design surface on which these decisions sit, see Sourcing Design Axes: Node and Client Ownership. For decomposition of work into nodes, see Service Chain Decomposition and Node Sourcing. For the two-horizon programme structure, see Sourcing Strategy Under Cost Pressure.

It exists because sourcing frameworks routinely establish that location is a consequence rather than a decision, and then leave practitioners without a rule they can apply to a specific piece of work on a specific Tuesday. The gap between a correct framework and a usable one is an instrument, and the absence of that instrument is why placement reverts to rate comparison under time pressure.

The method in four steps

The method is short enough to state in full before it is explained, and each step below is expanded in the section that follows it.

One. Establish that location is not the variable. The same supplier will frequently outperform an in-house team on one type of work and prove unable to hold another. That is not inconsistency; it is a staffing model behaving predictably. If the same supplier is capable and incapable at once, the supplier was never the variable being measured. What separates the two cases is whether the resolution path can be written down in advance, which yields three placement bands. See The discriminating question and Three bands.

Two. Recognise that a service estate holds three distinct supply types and usually buys them as one. Retained in-market delivery, owned remote centres and contracted supply are different purchases serving different purposes, and comparing them on a single rate ladder discards the property that distinguishes each. See Three supply types.

Three. Contract the drain rate rather than the expertise. Proficiency is not purchasable and no agreement transfers it. The rate at which people leave is countable, auditable and already reported, and it determines what a supplier can be asked to hold. See Contracting the drain rate.

Four. Replace the sourcing position with a constraint set and an evaluation step. A position is a fixed answer that begins decaying immediately, because the constraints determining it are set commercially and continue to move. One recorded set of what prevents work moving, and one point at which a proposal is scored against it before it becomes a decision. See The placement gate.

What this means in practice. The four steps are not a sequence to be completed. Steps one to three are readings of an estate that can be established in weeks; step four is a standing capability. An operation that completes the first three and skips the fourth has produced a sourcing position, which is the thing the method exists to avoid.

The discriminating question

Can someone write down, in advance, how this gets resolved?

Work for which the answer is yes is specified. Work for which the answer is no is discretionary — the person handling it determines the path.

What this means in practice. The test is not difficulty. A great deal of highly complex work is entirely predictable: it has many steps, several systems and a long handling time, but the sequence is known before the interaction begins. Conversely, work that appears trivial can be discretionary if resolution depends on judgement about a situation nobody anticipated.

The common error is to treat difficulty as the variable. Difficulty determines who is capable of doing the work. Specification determines whether capability can be built quickly or must be accumulated.

Why the question discriminates

Where the path is specified, competence is transferable through documentation and training, and a new person reaches the standard in weeks. Where it is not, competence is accumulated through exposure to cases, and time in the seat is the only mechanism that produces it.

This is why the same supplier will frequently outperform an in-house team on one work type and prove unable to hold another. The observation is a common one and it is routinely misread as inconsistent supplier performance. It is not. The supplier is performing exactly as their staffing model permits: strong where competence can be installed, weak where it must be grown.

Three bands

The question combined with customer contact produces three placement bands.

Band Work Rule
One Specified, no customer contact Place on cost. The measured penalty at the firm boundary is at or near zero for asynchronous non-customer-facing work
Two Specified, customer facing Externally eligible, with turnover contracted. Absent those terms, the saving being assumed by using a shorter-tenure pool is written into the business case explicitly rather than left implicit
Three Discretionary Retain internally. Tenure compounds here and cannot be purchased

What this means in practice. Most disputes about sourcing turn out to be disputes about which band a piece of work sits in, and that is a far more tractable argument than one about where the work should go. Two people who disagree about a destination but agree on the band have a solvable problem.

The band that carries the weakest evidence

Band three is the exposed one, and it should be stated as such. The literature supporting a customer-satisfaction penalty at the firm boundary finds that penalty statistically insignificant for services firms specifically, which is the sector most operations applying this method belong to.

Band three therefore does not rest on that literature. It rests on knowledge accumulation, which is independently supported: knowledge from learning by doing depreciates rapidly in service work and transfers within an ownership boundary but not across one.

The correct attack on the rule is that a supplier with contracted tenure could hold discretionary work. That proposition is testable and largely untested. If it holds, band three narrows and the method becomes cheaper to operate rather than collapsing.

Three supply types, one price ladder

A service estate of any size holds three structurally different supply types, and a common failure is to evaluate them against a single rate comparison.

Supply type What it is for Distinguishing property
Retained in-market delivery Work that cannot move for contractual, regulatory, clearance or proximity reasons Not a residual category. Its size is set by the constraint set, not by preference
Owned remote centres Work requiring accumulated competence at a lower cost base The only supply type that accumulates tenure. A supplier structurally cannot replicate this
Contracted supply Surge, ramp, a new language, work with a specified resolution path Speed to seat, which is genuinely purchasable and generally reliable

What this means in practice. Each type should appear as a separate line in a business case, because each is answering a different question. A supplier bought for unit cost is an owned centre bought badly. An owned centre sized for surge is contracted supply bought badly.

The most expensive of the three errors is treating retained delivery as whatever is left over, because it is the one category whose size is determined by constraints rather than by choice. Sizing it as a remainder means the constraint set has never been established, and the cost of the constraints is absorbed silently rather than funded.

Running owned centres primarily as the low-cost option is the second error, and it is subtler. It spends the only structural advantage that supply type has on something a supplier can already provide more cheaply.

Contracting the drain rate

Do not contract for proficiency. It is not purchasable and no contract transfers it.

What is being bought is a stock of accumulated experience held in the seats serving an account. The stock depreciates, and turnover is the rate at which it drains. The stock cannot be purchased. The rate can be constrained, and unlike proficiency it is countable, auditable and already reported in most agreements.

Three terms that can go into an agreement

Attrition ceilings measured on the named account, not on the site. Site-level attrition can hold steady while the account rotates underneath it. The account is the unit that carries the stock, and a site-level ceiling is satisfied by a supplier who is quietly reallocating people between clients.

Tenure mix floors rather than averages. A stated share of the named team above a stated service length. An average is satisfied by a small senior core while the remainder turns over completely, which is precisely the configuration the term exists to prevent.

Named team continuity. The term suppliers resist hardest, and the one that matters most, because rotation resets the knowledge stock regardless of what the headline attrition figure reports.

What this means in practice. These are not quality demands and should not be presented as such, because a supplier will price them as risk. They are the observable proxies for something already being bought and not currently specified. Framed that way they are considerably easier to agree.

The order-of-magnitude argument

A simple steady-state model, and it should be presented as a model rather than a measurement.

At thirty percent annual attrition, average time in the seat settles near three years. At ten percent, near ten.

Same price. Structurally different workforce. No rate card closes that gap, and the implication is that the attrition rate rather than the hourly rate determines what a supplier can be asked to hold. A supplier at thirty percent can hold bands one and two. The same supplier at the same price cannot hold band three.

Turnover is a design outcome, not only a supplier property

This is the part that changes who the argument applies to. Turnover in service work is driven substantially by job design rather than by contract type: roughly nine percent where roles are high-discretion and low-monitoring, against roughly thirty-six percent where they are low-discretion and high-monitoring.

A supplier asked to hold attrition down while being specified a low-discretion, high-monitoring job design has been given two incompatible instructions. Specifying discretion and monitoring is what makes an attrition ceiling achievable rather than adversarial, and it costs the supplier nothing to agree.

It also applies to internal delivery just as much. Stating it that way converts the section from an accusation about suppliers into a design argument, and removes the obvious counterattack.

The placement gate

A placement rule without an evaluation step is a document. The gate is the point at which a proposal is scored against known constraints before it becomes a decision.

It is not for the questions that already get asked

Where a commercial request reaches an operational leader who reaches a planning function, a sizing already comes back. The gate exists for commitments made without a question being asked at all, because the constraint was not recognised as a constraint. The characteristic cases are coverage floors — where a roster minimum exceeds any volume justification — language viability, and legality, none of which present as workforce questions at the point they are agreed.

Four outcomes

Outcome Meaning
Pass Eligible on every constraint. Should return fast, with the reasoning recorded
Fail Blocked by a hard constraint
Fail with a price Blocked by a soft constraint, with the cost of overriding it computed
Pass with a dated transition Eligible now, with a stated condition at which the placement is recomputed

The third outcome is the one most operations do not have, and it is the one that changes conversations. A proposal that fails on cost grounds remains a decision available to the business. It becomes a purchase rather than an assumption.

A gate that only ever subtracts is an obstacle and will be routed around. The clean-pass outcome is not a courtesy; it is what makes the instrument usable in an operation under delivery pressure.

The counterfactual, not the price

A frequent objection is that sizing already happens on request, which is usually true. What does not come back is the delta. Where a dedicated arrangement is requested, a headcount is produced; what the same work would have cost pooled is not. So the number reaching the commercial conversation is the cost of the team rather than the price of the request, and exclusivity is never visible as a separable cost.

Failure modes

Treating difficulty as the discriminator. Produces a rule that keeps complex-but-specified work internal at full cost, and pushes simple-but-discretionary work outward where it fails.

Contracting attrition without specifying job design. The ceiling becomes adversarial, is priced as risk, and is met by reallocating people rather than retaining them.

Measuring attrition at site level. Satisfied while the account rotates completely.

Building a gate that cannot say yes quickly. It will be bypassed, and the bypass becomes the process.

Publishing the bands without the evidence position on band three. The weakest claim then carries the most expensive delivery, and the first person to check finds it unaided.

Presenting the steady-state tenure arithmetic as a computation. It is a model. Stated as a measurement it invites an argument about the model instead of about the workforce.

Maturity Model considerations

An operation at L1–L2 typically places work by precedent and rate, with no recorded constraint set and no evaluation step. Attrition appears in agreements as a target without teeth.

At L3, bands are recognised and applied inconsistently, and attrition terms exist at site level.

At L4, placement runs against a recorded constraint set with a gate, attrition is contracted at account level with a mix floor, and the counterfactual cost of exclusivity is computed as a matter of course.

At L5, the constraint set is maintained continuously, placements carry reversion conditions, and the gate recomputes when a constraint moves rather than when someone remembers to ask.

Use this with Claude

A ready-to-deploy instruction set and reference files for building the placement engine are at Wiki:Packs/Placement Decision Engine Build (CP-WFM-008) — building the register, gate, cards and recompute log.

See Also

Retrieved from "https://wiki.wfmlabs.org/w/index.php?title=Placement_Rules_and_the_Tenure_Contract&oldid=3918"