Migration Modeling Readiness

From WFM Labs
Six sections, answered against open, with the staffing effect of what remains open.

Migration modeling readiness is an instrument run before a service migration is modeled. It asks a fixed set of questions whose answers determine what the plan may assume, and converts every unanswered question into a stated row on The Assumption Register carrying an owner, a date and a sized effect.

Its purpose is not to prevent modeling until everything is known. Most migrations are modeled under substantial uncertainty and must be. The instrument exists so that the uncertainty is discovered before the plan rather than after it, and so that the parts of it that move the answer are separated from the parts that do not.

Why a fixed instrument rather than judgment

A capacity model is a chain of inherited values — contact rates, handle times, channel splits, seasonal shape, shrinkage — each carried forward on the implicit assumption that what it measures has not changed. An experienced planner will interrogate the values that experience has taught them to distrust. The failures that recur are in the values nobody thought to interrogate, which are systematically those embedded in a measurement definition rather than in a number.[1]

A fixed list is therefore not bureaucracy substituting for expertise. It is the mechanism by which a defect found expensively on one migration becomes a question every later migration has to answer.

The six sections

Items are mandatory by archetype; an item that does not apply is marked as such rather than left blank.

A · Point of departure

  1. Do we hold the departing operation's own contact data for the markets being moved — not a platform norm, and not a whole-account average?
  2. Which periods are pre-migration, and have later periods been excluded from the baseline?
  3. What is the transaction definition on each platform, and do they match?
  4. What is the contact definition — offered, handled, or answered within target?

Item A1 is the most frequently failed item in the set, and its failure is not visible until volume arrives. The rule it enforces is set out at Migration Archetypes.

B · Demand build

  1. Was annualized demand built from transactions or from contacts, and which is treated as the driver?
  2. Is the seasonal profile the client's own, or a house profile applied to them?
  3. Does the client's weekly profile match the house profile, and has anyone checked?
  4. Is there an intraday profile, and is it theirs?
  5. Market split — do we have one, and at what granularity?

B3 and B4 are cheap to check and routinely skipped. A client whose weekly shape differs from the house profile will break an otherwise sound plan at the day level while the weekly total looks correct.

C · Composition and channel

  1. Channel split today, and expected after?
  2. Is any channel new, or is any existing channel's synchronicity changing?
  3. Are entry points changing — integrations, self-service surfaces, automated agents?
  4. Is deflection or containment assumed, and on what evidence?

C3 catches the case in which the customer's route changes while nothing the agent sees changes. Every agent-side measure remains stable and the volume effect arrives without warning.

D · Experience change

  1. What does the customer meet that is different?
  2. What does the agent meet that is different?
  3. Timeouts, session rules, and whether conversation history carries across?
  4. Is there a handle-time penalty at cutover — what size, and what decay profile?

Where the channel is genuinely new there is no series to extrapolate from and the forecast becomes a judgmental one supported by analogy.[2] D4 is distinct from a ramp: the performance loss associated with moving knowledge into a changed context applies to experienced staff as well as new ones.[3]

E · Supply

  1. Training length, nesting and the proficiency curve — whose numbers, measured or assumed?
  2. Shrinkage — measured or assumed?
  3. Concurrency, and is the handle-time measure elapsed session time or agent working time?
  4. Eligibility constraints — language, clearance, location, contractual?
  5. Knowledge loss on exit, and what carries it?

E3 is a single question that can change a workload calculation by the concurrency factor, and it is rarely recorded because the choice is not experienced as a choice.

F · Data and governance

  1. Which figures can actually be measured weekly, and which will the platform not report?
  2. Where do two metrics share one name across the platforms in scope?
  3. Who owns each open item, and by when?
  4. Has the incoming plan been laid against the outgoing plan, and who signed it?

F4 is the only pass/fail item. Every other item may remain open with an owner and a date. F4 is a governance step rather than an input, and its absence is the failure behind most of the others: a defect in the plan that nobody compared to the operation being replaced is a defect nobody was positioned to find.

How the output is used

Each item resolves to one of four states, and the state determines what happens next:

State What it produces
Answered, evidenced A value in the model, graded Established, with its source recorded
Answered, asserted A value in the model, graded Asserted, and a measure named that would confirm it
Open A register row with an owner, a date, and the staffing effect of the range it could take
Not applicable Marked, with the archetype that excludes it

The instrument is run once at kickoff and reviewed at each phase gate. Between gates, open items appear on the assumptions artifact of the weekly pack described at The Migration Health Pack until they close. Ranking open items by the product of their staffing effect and the weakness of their evidence produces the analytical work queue directly.

Failure modes

  • Run after the model. The instrument then documents a plan rather than shaping one, and the items it would have caught are already embedded.
  • Treated as a gate rather than a discovery. Refusing to model until every item is answered guarantees the instrument is bypassed. The design intent is the opposite: model with open items, but sized and owned.
  • Items answered by the modeler alone. Several items — platform reporting limits, contractual eligibility, transfer law — are not knowable inside the planning function, and an answer invented there is worse than an open row.
  • Open items without a staffing effect. The list becomes a set of disclaimers and stops directing work.
  • F4 graded rather than passed. It is the one item with no partial state.

Maturity Model Position

At Level 1–2 the questions are asked informally, by whoever has been burned before. At Level 3 a fixed instrument exists and is run at kickoff. At Level 4 every open item carries a computed staffing effect and a named external owner, and the instrument is reviewed at each phase gate. At Level 5 the item set is revised from the organization's own record of prior migrations, so that each failure adds a question rather than a lesson.

See Also

References

  1. ↑ Morgan, M. G., & Henrion, M. (1990). Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis. Cambridge: Cambridge University Press. doi:10.1017/CBO9780511840609. Treats model structure, not only parameter values, as a source of uncertainty requiring explicit treatment.
  2. ↑ Hyndman, R. J., & Athanasopoulos, G. (2021). Forecasting: Principles and Practice, 3rd ed. Melbourne: OTexts. The chapter on judgmental forecasting sets out the conditions under which no relevant historical series exists.
  3. ↑ Argote, L. (2013). Organizational Learning: Creating, Retaining and Transferring Knowledge, 2nd ed. New York: Springer. doi:10.1007/978-1-4614-5251-5.
Retrieved from "https://wiki.wfmlabs.org/w/index.php?title=Migration_Modeling_Readiness&oldid=5093"