Planning scope and handling change
10 min read
Lesson objectives
- Translate scope into an assignment plan with deliverables, evidence access, risks and change control. Use the engagement work plan to produce a assignment charter and verify dependencies and acceptance are testable.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
After the sponsor and specialist agree on a question, a team may think execution can begin. But “analyze resolution time” says nothing about when records arrive, what the first review contains, who checks the metric definition, or what happens if half the records are unusable. An assignment plan makes the promise executable. It need not be long. It must show the link among input, output, dependency and acceptance. The ICMCI competence framework describes consulting assignments across contracting, execution and closure, with clarity of role and agreed deliverables. The plan below is a teaching tool for this case, not an ISO form imposed on every client.
Work backward from the decision. If the final choice will compare hiring, classification repair and shift redistribution, a trade-off table cannot be credible before the team defines time, tests record quality, estimates rework and learns the budget constraint. Write a delivery chain: confirm data definitions; record hypotheses and collection needs; present a provisional diagnosis with uncertainties; develop options that can be trialed; deliver a decision brief and working files. Every checkpoint has a tangible output and an authorized reviewer. “Finish the analysis” is not a useful checkpoint on its own. What exactly can the client open and test at its end?
For the fictional service center, the first week could run as follows. On day one the sponsor confirms the decision question and the process owner confirms the two-site and case-class boundary. On day two the data steward supplies a field dictionary and a privacy-protected sample of 30 tickets. On day three the specialist delivers a metric-definition sheet with a list of records whose IDs cannot be linked; data and process owners review it. Only after that review is the quarter-to-quarter comparison calculated. On day five the specialist presents capacity, rework and measurement-error hypotheses, with evidence for or against each. These dates are fictional and depend on access; they are not a universal timetable.
Acceptance is more than a polite sign-off. The metric sheet passes when opening and closure events are defined, the ticket population is specified, excluded records are counted with reasons, and another person can recalculate five selected cases. If the five-case check fails, do not carry the figure into a decision brief. The options brief needs a different test: approximate cost under stated assumptions, likely effect with uncertainty, quality risk, and a pilot decision owner. Attractive slides are not an acceptance criterion for analysis. The criteria let the client find an omission before it hardens into an influential recommendation.
One can make a four-column plan: deliverable, required input, reviewer, and acceptance test. The metric-definition row requires a field dictionary and linked-ID sample; the data steward and process owner check five recalculations. The diagnostic row requires usable records, rosters and classification rules; the process owner checks whether each hypothesis has supporting or opposing evidence and a stated limit. The recommendation row requires a qualified diagnosis and cost bounds; the decision owner checks time, quality and cost trade-offs, including why an option was rejected or deferred. These rows show why raw data cannot responsibly become a hiring recommendation in a day.
Record dependencies and owners because a missing input can change the truth of an inference. Without rosters it is hard to test capacity. Without old classification rules, rising reopen rates cannot be connected confidently to the rule change. Give each dependency a needed-by date, provider, limited fallback if unavailable, and consequence for confidence. For example, an absent roster does not justify assuming stable capacity. It permits a description of the observed trend and an explicit statement that the capacity explanation remains unresolved. That is defensible; hiding the absence in a footnote is not.
After approval, management asks to add a third center “to make the report comprehensive.” The request may alter the population, source systems, interview count and privacy conditions. The specialist should neither reject it automatically nor absorb it silently. A short change record states the proposed addition, why it matters to the decision, expected effects on time, evidence, access and output, and the options requiring an authorized choice. If the third center uses a different ticket system, combining it in one number before aligning definitions may weaken the report. A preliminary check, a later phase, or an approved change to scope and date are possible responses.
Here, records from two centers support a preliminary diagnosis in ten working days. The third center lacks a persistent ticket ID and could need another week of linkage work. Put two choices to the sponsor. Keep the first decision confined to two centers and state the third as a limit on generalization; or formally add it and move the recommendation date after linkage is checked. If the decision is to roll out a solution to every center immediately, exclusion may be unacceptable. Say so rather than deliver a broad recommendation on narrow evidence. If the decision is a two-center pilot, the narrower work may proceed. This is a trade-off between a timely bounded decision and wider evidence later, not a test of which team is “helpful.”
The plan also names risks that can halt work. If the ticket extract contains personal details irrelevant to the comparison, request de-identified fields rather than a full copy. If policy prevents even that, ask the authorized owner to assess an aggregate alternative; the specialist should not copy data onto a personal device. If the sponsor is invested in proving the hiring case, keep hiring among the tested options and use declared criteria. Objectivity takes a method, not just a good intention. If a question requires legal or information-security judgment outside the specialist's competence, refer that part to the relevant expert rather than claiming authority.
Lesson summary
A consulting assignment starts with a decision question, not the sponsor’s preferred remedy