From a sponsor request to a testable consulting mandate

The chief executive of a fictional consumer lender says, ‘Give me one number proving that expansion is safe before Thursday’s investment meeting.’ This sounds specific but combines three different matters: the expansion decision, the firm’s exposure to losses it cannot absorb, and the way that exposure is summarized. A single number cannot decide for the board. The consultant first turns the request into a testable decision: which expansion plan, at what scale and horizon, subject to which liquidity and capital limits? Without a comparator, even an accurate number has little decision meaning. The real choice may be phased expansion rather than a simple yes or no. Clarifying that choice does not disregard the sponsor; it makes the evidence useful to the decision maker.

Identify the sponsor, the decision owner and affected users separately. The sponsor commissions the work and grants access, while the authorized committee may decide the investment. Credit, treasury and operations staff each hold pieces of the record; their incentives and definitions may differ. Specify the evidence scope: the existing loan portfolio, delinquency under a stated ageing rule, cash collections, funding for expansion, and customer or geographic concentration. Specify the exclusions too: no certificate that all risks are ‘safe’, no certain default forecast, and no legal, assurance or investment opinion beyond the mandate. These boundaries prevent an analytical report from becoming an unsupported guarantee.

Form hypotheses that evidence could disprove before you gather the files. Rising arrears could reflect weaker new underwriting, a downturn among customers, or a changed day-counting rule after a system migration. Each explanation predicts different observations and has a possible refutation. If underwriting deteriorated, recent vintages should perform worse after allowing for time on book. If migration caused the signal, there may be a jump on the conversion date without a matching change in cash collections. These are not findings in advance. They are a plan against selecting evidence that merely confirms the sponsor’s view. Record periods, denominators, definitions and the owner of each file before calculating rates.

Connect the scope to objectives. ISO’s public overview of ISO 31000:2018 stresses integration of risk management with strategy and activity, responsibilities and resources; COSO places enterprise risk management alongside strategy setting and performance. Citing those overviews does not mean claiming to implement either complete framework. Here the objective is profitable loan growth that the firm can fund. Assess the possibility of additional defaults, potential loss size, required cash and the return from growth together. A delinquency rate alone conceals the trade-off the committee must decide. Choose a small set of measures that illuminate that trade-off and explicitly mark any measure unavailable from the records.

Test feasibility with three questions. Can the data be accessed under clear authority and suitable protection? Is there enough time to reconcile definitions and totals? Can the decision owner accept a conditional recommendation or a delay if the evidence is thin? If Thursday is too early for a vintage test, call the result a preliminary review and identify a lower-exposure interim decision, such as a capped first release subject to a check two weeks later. Deadline pressure does not create certainty. Equally, a limited gap need not prohibit every recommendation; separate what is known now from what is required before the next increase.

The resulting engagement note is brief but operational: decision purpose, alternatives, period, units of analysis, indicator definitions, sources and access rights, interview schedule, evidence-quality criterion, reporting format, opinion boundary and a process for changing scope. Ask the sponsor to agree to the question and change procedure, not to a preferred result. The consultant may conclude that some exposures cannot be measured from these data. That is more useful than compressing disparate scores into one figure that hides differences among event likelihood, impact size and speed of onset.

Check the mandate before interviews: does the sponsor want an independent conclusion or language supporting a decision already made? If the request is ‘prove expansion is safe’, require the scope to permit a finding against expansion, direct questions to treasury and credit, and resolution of conflicting figures. If delinquency data are withheld while only sales files are shared, the result cannot be described as a complete risk assessment. A bounded sales-opportunity review is possible, or the risk opinion must await the missing record. Define confidentiality as well: who may see customer-level data, and where will aggregates suffice? End the scope with a decision the report could genuinely change, such as cutting a proposed 15-million release to five million if a delayed-collection case breaches the cash floor. The board can then judge the engagement against a real question rather than a vague promise of reassurance.

An engagement note should define evidence completeness, not merely list files. Delivery of a loan export is insufficient until closing balances reconcile to the ledger within an explained difference and post-cut-off changes are identifiable. A 2% mismatch is not automatically minor if it sits in large or overdue loans. Agree a data-request closing date and what happens if material information appears afterward. Issue a dated supplement instead of silently rewriting an earlier report. The sponsor can then see what work was commissioned, while decision makers can distinguish a completed review from a preliminary estimate constrained by time or access.