The Migration Health Pack

Requirement and supply drawn on one axis, with reinforcement arrival marked. Drawn separately, the two are a headcount chart and a service chart; drawn together, the timing gap is visible.

The migration health pack is a fixed set of six artifacts, refreshed weekly, reporting the supply and demand position of every migration on an organization's near horizon in a single repeated shape. It is a governance artifact rather than a model: the models sit behind it, one per migration, and the pack is the fixed surface they report through. It is applied to a portfolio, and its value comes from covering every migration on the horizon in one shape rather than from documenting one migration thoroughly.

The problem it addresses

Organizations carrying several concurrent transitions tend toward a characteristic pattern. Each migration is planned by whoever is closest to it, in whatever form that person favors, and reported when it becomes urgent. Attention is allocated by escalation, so a migration drifting quietly attracts none until it is no longer quiet. And because no two reports share a shape, each review begins with an orientation, and the questions asked of one migration are not asked of the others.

The compounding effect is that a defect discovered expensively on one migration remains knowledge held by the people who lived through it, and the next migration repeats it because nothing in the reporting form would have surfaced it. Knowledge held by individuals rather than embedded in a routine is the form of organizational knowledge least likely to transfer between units. A fixed pack addresses that transfer problem structurally: a defect found once becomes a field every subsequent migration must complete.

The six artifacts

1. Approach

How the plan was built: the archetype declared (see Migration Archetypes), the walk from point of departure to point of arrival with the adjustment and its basis recorded at every step, and a short list of what was deliberately not modeled, with reasons.

The artifact exists because an unstated approach is an unchallengeable one. It is the artifact most often absent and the one most likely to have prevented the failure reported in the others. Its construction is treated at Migrating a Book of Business.

2. Assumptions

The migration's assumption register — value, grade, sensitivity, owner and date per row, with the mandatory rows set by the archetype. The artifact is described at The Assumption Register.

Where root-cause or quality work runs alongside the migration, this is the artifact it reports into.

3. Requirement against supply

The forward position: requirement and available supply on a common time axis, with the projected service outcome per period, and a decision band recording any decision currently required, the date it is needed by, and what moves if it slips. The band is empty in most weeks; its value is that when it is not empty, the decision carries a date.

Two design conventions carry most of this artifact's value.

Supply arrival is marked on the same axis as requirement. Where the reinforcement plan is presented separately from the requirement outlook, a wave landing after the peak it was intended to cover is arithmetically visible and perceptually invisible, because the reader has to hold two charts against one another to see it.

The service projection rather than the staffing surplus is treated as the risk signal. The relationship between staffing and service is strongly non-linear in the neighborhood of the offered load, a consequence of the convexity of the delay and abandonment functions described at Erlang C: a small deficit produces a disproportionate service loss, while a small surplus produces very little service gain. A table read on the surplus column therefore treats a marginal deficit as a rounding error when it is a material service miss.

4. Mitigation

The supply mobilization plan: each source of staff, its volume, its training and nesting duration, its owner, and the dates at which it begins, enters supervised production and becomes fully productive.

The artifact carries one convention beyond a schedule: a compression or acceleration is recorded together with its cost. Shortening a training or nesting period to recover a near-term service position is a legitimate decision, but it purchases the near term with proficiency in a later one, and the exchange is only a decision where the second half is recorded. Where the proficiency penalty can be expressed through the plan's own handle-time sensitivity, expressing it that way makes the two halves comparable. The mechanics of the nesting period are treated at Agent Onboarding and Nesting Period Management and the recovery profile at Speed to proficiency curve.

5. Performance

Departure and arrival measured side by side, per channel, on the same definitions: offered, handled, answered within the service target, service level, average speed of answer, abandonment and handle time.

Reporting both sides is a protection as well as a courtesy. An arriving operation reported alone reads as a sequence of failures; reported against the operation feeding it, the same figures show where the volume originated, whether the departing side is being over-served at the arriving side's expense, and what a like-for-like comparison actually supports.

Two conventions are worth stating. Handled volume is reported rather than volume answered within the target, because publishing the in-target count in a column labeled as handled understates workload by the miss rate — which is largest precisely when the operation is performing worst. And events are annotated on the day they occur: outages, releases, external disruption, training draw-down. Annotation after the fact cannot be distinguished by the reader from explanation constructed after the fact, which is why the convention is same-day.

Metrics the reporting platform does not produce are not omitted from the artifact. They are shown as unresolved, recorded as an open row on artifact 2 with an owner, and noted on the definitions page as a known gap in comparability.

6. Quality

The quality position on a fixed lookback window and a single intent taxonomy, reported as a trend with the small number of dominant failure reasons beneath it. A lookback that moves with the reporting date produces changes that are artifacts of the window rather than of the operation.

Where a quality finding has a capacity consequence it does not remain on this artifact; it becomes a row on artifact 2 with a sensitivity attached.

The portfolio companion: a definitions page

Alongside the six, one further record is held once for the portfolio rather than per migration, and is not itself one of the six: a definitions page recording every metric, its source, its exact definition, and each place it is not comparable across the platforms in scope. It is held centrally because comparability problems are properties of the estate rather than of any one migration, and holding it per migration causes the same comparability argument to be relitigated in every review. See The Definitions Register.

Cadence and forum

The pack is refreshed weekly and reviewed in one forum covering every migration on the near horizon, in a fixed order.

The forum's most common failure is not analytical. Organizations running several transitions already carry a dense meeting calendar, and meeting load is a frequently reported constraint on managerial attention. A review added to a saturated calendar competes for the attention of exactly the people whose attendance makes it useful. Substituting the review for an existing forum, and naming the substitution when the review is proposed, addresses the constraint directly; adding it does not.

The pack's completion test is behavioral rather than formal. It is finished when it answers, consistently from week to week, the large majority of the questions that recur, and when each further migration entering the roadmap can be slotted into the same shape without redesign.

Failure modes

• The pack becomes a per-migration document. Written once per transition in that transition's own shape, it loses the comparability that was its purpose.
• Artifact 1 is skipped because the model is already built. The approach artifact is what surfaces a wrong baseline, and it is omitted precisely where it would have been most useful.
• Requirement and mobilization remain on separate axes. This is the most common route to a reinforcement plan arriving late without the lateness being noticed.
• Quality findings are reported and never converted. A quality finding with a capacity consequence that does not become an assumption row is a fact the plan does not hold.
• The forum is added rather than substituted. Attendance decays and the pack survives as a document that is produced but not read.

Maturity Model Position

At Level 1–2, migrations are reported on escalation, in whatever form the owner prefers. At Level 3, a fixed pack exists and is refreshed on a cadence, covering the migrations currently causing concern. At Level 4, the pack covers the full near-horizon portfolio whether or not a migration is causing concern, carries sensitivities, and its assumption artifact is the defined destination for analytical findings. At Level 5, the pack's forward position feeds migration sequencing decisions, so that concurrent high-uncertainty transitions are not committed against a single planning capability.

See Also

• Migration Archetypes
• The Assumption Register
• Migrating a Book of Business
• Asynchronous-to-Synchronous Channel Conversion
• Capacity Planning Cycle
• Agent Onboarding and Nesting Period Management
• Executive Issue Register
• Answer-First Reporting
• Service Level During Work Migration
• Migration Modeling Readiness
• The Interim Data Layer