Consulting assignment and boundaries · Lesson 1

Turning a request into a consulting question

9 min read

Lesson objectives

  • Convert an ambiguous sponsor request into a bounded decision question and acceptance test. Use the problem statement and scope card to produce a scoping brief and verify sponsor and process owner can agree on metric, population, decision and exclusions.
Reading tools

The service director says, “We are slow because the team is too small; recommend four new hires.” The request sounds precise because it contains a number. Yet what is slow: first response, resolution, or delivery? An average may also conceal a narrow tail of long cases. If a specialist starts by costing four hires, the analysis has already accepted a cause that has not been tested. The first consulting skill is to turn the sponsor’s preferred answer into a decision question: “What is driving the rise in ticket resolution time, and which intervention can reduce it within cost and quality limits?” That question leaves staffing open as one possible explanation.

An assignment is an agreement about what to examine, produce, and deliver to whom. It is not open authority to change the client’s operations. ISO 20700:2017 is presented by ISO as guidance on the delivery of management consultancy services. It is neither a licence conferred by this course nor a universal law. The ICMCI competence framework places understanding the client business and stakeholders in a consultant’s capability. We use those sources to orient scope practice, while the specific card below is an instructional tool of this course; we do not attribute its fields to a full standard we have not reviewed. Distinguish the client organization, the sponsor requesting the work, the decision maker, and people who will live with the outcome. One person can fill more than one role, and their interests can still differ.

Start with the decision, not with whatever data is easy to obtain. Ask what decision the report will inform: hiring, a workflow pilot, a change in priorities, or a decision to leave the process alone. Ask what would make the decision acceptable: a resolution-time threshold, a cost ceiling, and a quality floor. If the sponsor lacks final authority, record who can approve the decision. Define the unit being measured. A “closed ticket” is not the same as a customer or an email. State the start and end timestamps, period, and case types. Two figures with the same label may be incomparable when their denominators or boundaries differ.

A useful scope card separates five things. First, state the observed problem in testable language rather than assigning blame. Second, write the decision question and plausible alternatives without choosing among them. Third, name the organizational and time boundaries: which sites, ticket classes and dates. Fourth, specify outputs that serve the decision: a metric definition, a traceable data table, an options comparison and a recommendation with assumptions. Fifth, state authority and exclusions: who may grant record access, how customer data will be protected, and which steps require separate approval, such as reading sensitive messages, changing a service policy or hiring. “A report” alone is hard to accept or reject. “Compare three interventions on resolution time, cost and quality risk” has a usable acceptance test.

Consider a fictional company with two service centers. It closed 1,200 tickets last quarter, with a median opening-to-closure time of 2.1 days. This quarter it closed 1,440 tickets and the reported median rose to 3.0 days. The director says headcount was unchanged and asks for four hires. Operations says the reopened share rose from 8% to 15% after a classification change. A data steward says some tickets were administratively closed and reopened under new identifiers. The evidence conflicts. Higher volume could strain capacity, rework could consume time, and identifier changes could corrupt the comparison. None of those observations alone proves that hiring or classification is the cause.

My initial scope card states that the decision is to select a first action to reduce resolution time without raising reopen rates or exceeding a client-defined budget. The examination covers routine tickets at both centers over the two quarters, while separating urgent cases and cases whose identifiers changed. Resolution runs from the first valid opening to a stable closure after a follow-up window agreed with the process owner; first-response time is tracked separately. Inputs include ticket records, old and new classification rules, rosters and a privacy-protected sample of reopened cases. The output compares a capacity explanation, a rework explanation and a measurement-error explanation before proposing testable interventions. The assignment does not authorize hiring or viewing sensitive content without permission.

Test the card with the decision maker and the data owner. The decision maker may say, “I need the staffing decision this week.” A rapid brief can offer a conditional estimate, but the reopen and identifier issues make a categorical cause claim unsound unless a sample can separate them. Offer a sequence: in two days, check the metric definition and 30 tickets spread across sites and periods; then decide whether the data permit a comparison of capacity with rework. Thirty is a fictional initial check size, not a statistically guaranteed sample. If access fails, narrow the scope and lower the confidence of the conclusion rather than inventing an answer.

“What happened?”, “why?”, and “what should we do?” are different questions. A median of 3.0 rather than 2.1 days describes a change; it does not establish a cause. Case mix and the closure definition might explain part of it. Only after validating the measure should the specialist test competing mechanisms. Only then does a comparison among staffing, classification improvement and workload allocation become useful. The starting card need not list every analysis one could perform. It needs a chain that ties each piece of work to a decision. Record stopping and escalation points as well: no legitimate access to records, unresolved disagreement over the metric, or a request to use individual employee data for an appraisal outside scope. Those are authority issues, not arithmetic problems.

Review the card with two counterfactual questions. If corrected identifiers show no rise in resolution time, would hiring four people still follow from this assignment? If the answer is “yes regardless of evidence,” the client is commissioning an implementation study of an already chosen option, not an open diagnosis. That can be a valid assignment if stated honestly. Second, if the rise is real, does the card collect enough evidence to distinguish capacity from rework? Without rosters, reopened cases and case mix, it does not. These tests prevent the engagement from producing slides that merely restate the sponsor’s opening belief.

Lesson summary

A consulting assignment starts with a decision question, not the sponsor’s preferred remedy