Chapter 4: Exercise Ethical Judgment and Tailor Deliberately

The forty-truck offer

Three days into the flood response, Priya Raman, Northstar’s procurement officer, puts a one-page offer on the table. Fastlane Freight will put forty trucks on the northern route within 48 hours at a rate 60 percent below the next bidder, on one condition: half the estimated contract value up front, about 900,000 units, paid before the first load. The northern hub is 16 tons a day short of what 14,000 shelter residents and six district health posts need. Medicines at the hub will last four days. The emergency grant’s spending window closes in ten days.

Farida Aziz, the compliance and safeguarding adviser, has already run the quick checks. The firm registered three months ago. The fleet it declares does not match its tax filings. It has no safety record. Two of its named drivers hold suspended licenses. Full due diligence takes five to seven working days, and the registries are as strained as the roads.

Mateo Herrera, the field logistics coordinator, is blunt. “The hub is rationing insulin. Every day we do not move, someone who could have been treated is not.” Ines Carvalho, the donor liaison, is not pressuring anyone; she is stating a fact. The window closes on the tenth day, and unspent funds lapse. Re-programming the grant is possible but takes three to four weeks and is not guaranteed.

Nobody in the room wants to delay relief or hand 900,000 units to an unverifiable firm. The temptation is to dissolve the tension instead of deciding it: sign the contract, skip the checks, verify them after the first run. Daniel Okafor, the relief response lead, has until evening to decide. This is tailoring under pressure: choosing, with incomplete information, what to keep, what to compress, and what to escalate.

The floor below compliance

Accountability has two layers. Compliance means the process was followed: the checklist was signed, the gate was passed. Accountability means the consequences are owned: what happened to the people the work was supposed to serve, who could have known. The second layer is where ethical judgment lives, because the rules never cover every situation; in an emergency they cover fewer: what to do with a verified supplier, not what to do when verification itself is the delay.

Stewardship is the working attitude for that gap. You are the temporary custodian of other people’s money, information, and trust, and your duty runs to the people the work affects, not only to those who approved it. When the rules are silent or too slow, the decision is still yours, and it is still yours when the consequences arrive. Formally compliant projects can fail ethically: the paperwork was perfect and the people were harmed.

The practical instrument is the ethical decision record: a short written entry made when a decision involves pressure, a conflict of interest, or the temptation to cut an obligation. Seven elements: the context and the pressure; the decision and the options considered, including the tempting one; who decided and who else was consulted; what evidence existed and how strong it was; what was preserved and what was simplified or deferred; what would change the decision, with a revisit date; and who was told, with an open path to challenge. The record is not bureaucracy. It is the discipline of making the decision twice, once in the room and once on paper where it can be examined. The conflicts that trigger it are rarely dramatic: the evaluator grading work she will profit from, the supplier who is a director’s cousin, the funder whose contribution pays the reviewer. They are ordinary, and naming them is the defense; the record makes them visible instead of ambient.

The ethics of bad news

Forecasting is an ethical act before it is a technical one, because every forecast moves other people’s decisions. A delivery date is not a number; it is a claim sponsors, clinicians, donors, and communities will rely on. Honesty about what is known and what is not is the ethical content: a range, an assumption, a confidence statement, an owner. Chapter 15 builds the mechanics of estimation; the foundation is here.

The asymmetry of bad news is brutal and simple. Said early, bad news is information: the recipient has options, can re-plan, can fund, can change scope. Said late, the same fact is a betrayal: the options are gone, and the only remaining decision is who gets blamed. Every week of concealment transfers risk from the person who hid the truth to the people who relied on the report, without their consent. That is not a communication failure; it is a transfer of cost. The leader who waits until a red dashboard is forced out has already made everyone else’s decisions for them.

An honest forecast is not pessimism. It is the difference between “the date holds if two risky assumptions hold” and “the date holds,” stated so a decision maker can hear the first sentence and ask about the second. Chapter 2’s green dashboard is the mirror of that failure: every row green, with five success domains behind it.

Why good people ship green

The pressures that produce dishonest reporting are rarely crude. Nobody says “lie to me.” They say: the board needs a green quarter, the partner needs a renewal signal, the funder needs a spend rate, the team needs a win. Each is a legitimate interest. Each, left unexamined, becomes a reason to make the numbers say what the situation does not.

The classic instruments of managed truth are four: percent complete with no measurement rule, always arguable to 80; a rebaseline that quietly erases what the old plan taught; a launch date moved without a scope change, so the date stays “committed” and the work simply gets faster; a demonstration that passes the friendly cases and never the hard ones. They are the slow conversion of reporting from evidence into negotiation, and the tell is consistent: the metric and the lived situation disagree, and nobody updates the metric.

The mechanism has a name. The economist Charles Goodhart observed in 1975 that a statistical regularity collapses when pressure is placed on it for control purposes; Marilyn Strathern later compressed the idea into the line most people quote: “When a measure becomes a target, it ceases to be a good measure.” The project version is blunt: a metric people are rewarded or punished for will be managed, and the management is usually invisible in the metric itself. The counter-measure is not more oversight. It is keeping the measure attached to an observable reality, a count in the field, a delivery receipt, a working cold chain, and at least one metric the people under pressure cannot adjust.

The five things you do not price

Some obligations cannot be traded for speed, and the emergency is exactly when they must be named, because the pressure to price them is strongest.

Safety. The project must not put people at risk: workers, drivers, staff, volunteers, the public, and the people the work serves. At Northstar the floor is concrete: a suspended license on a flooded road endangers the driver, the convoy, and the community it passes.

Privacy. The project must protect information about people. Relief registration data is the extreme case: names, ages, needs, locations of people already displaced and vulnerable. Published carelessly, a shelter list becomes a targeting list. Access must be named, limited, and justified, and bulk requests, including from well-meaning authorities, get a protection assessment before they get data.

Accessibility. The work must reach everyone it claims to serve: people with disabilities, older people, those who cannot travel to a distribution point. A response that serves only the mobile half is a design flaw, not half a response.

Fairness. The project must not extract value from the people it touches: local workers paid and treated decently, local suppliers and markets not wrecked by an operation that undercuts them, promises to communities kept.

Environmental responsibility. Relief operations consume fuel, produce waste, and can damage a region already damaged; disposal of expired medicines and packaging is not an afterthought.

The humanitarian sector gives these floors a shape. The do-no-harm principle, developed by Mary B. Anderson and colleagues through case-based research and published in 1999, holds that aid must be assessed for the harm it can cause, not only the good it intends. The Sphere Handbook (Sphere Association, 2018) sets out the sector’s shared minimum standards for water, sanitation, food, shelter, and health; Northstar operates against them by choice and by contract. The discipline for the project leader is simpler to state: when a practice protects a person, the form may be lightened but the function is not negotiable, and no expected-value calculation licenses its removal.

Tailor, do not delete blindly

Tailoring is the deliberate choice of how much practice a project needs: which events, artifacts, roles, and controls exist, in what form, at what cadence. It is the art of the minimum, and it belongs in this chapter because it is where ethics and method meet. A project can fail by excess, drowning in ceremony no one reads, or by deficit, stripped so bare that nothing protects anyone. Both are tailoring failures; only the second is usually discussed.

This chapter feeds the Judgment term of Project Mastery = Judgment × Alignment × Delivery × Learning: judgment is only sound under an ethical floor, and tailoring is judgment applied to practice. What tailoring should depend on, in a phrase: consequence, complexity, uncertainty, novelty, and capability. Consequence: higher stakes mean stronger controls and less delegation without guardrails. Complexity, from chapter 3: short feedback loops, small probes, and decision rights at the point of feedback, not thicker plans. Uncertainty: ranges, options, and revisit dates rather than precise commitments. Novelty: more rehearsal, expert review, and verification for situations new to this team. Capability: the practices a team can run lightly versus the practices that would be theater in its hands. The five factors rarely agree, and that is the point.

The decision tree is the working instrument. Every practice you are tempted to drop goes through four gates in order.

Figure 4.1: Tailor, do not delete blindly. Every practice passes
through four gates before it may be lightened or dropped.

  a practice you are tempted to cut
        |
        v
  required by law, regulation,
  contract, or a safety standard?  ---- yes --> keep as required;
        |                                      escalate, never cut
        no
        v
  protects a person from harm
  (beneficiary, staff, public)?    ---- yes --> simplify the form,
        |                                      keep the function
        no
        v
  improves a decision you actually
  make, or transfers accountability? - yes --> minimum viable form,
        |                                      with owner and review date
        no
        v
  ceremony: cut it, record why,
  name what replaces it

The gates are load-bearing in order. What law, regulation, contract, or a safety standard requires stays required: escalate rather than cut if you cannot meet it. What protects a person keeps its function in a lighter form. What improves a decision you actually make, or transfers accountability to a named owner, keeps a minimum viable form with an owner and a review date. What serves none of these is ceremony: cut it, record why, and name the replacement, even when the replacement is “nothing, and here is why.”

Minimum viable governance is the lightest arrangement that still produces a decision, an accountability, or a protection. Its opposite is not “no governance”; it is unexamined governance, form without function. The detection questions are simple. Who reads this? Which decision changes because of it? Who is protected by it? What happens when it is late? If the answers are nobody, none, nobody, and nothing, it is theater: retire it.

The delivery approach changes how tailoring happens, not whether. In predictive delivery, tailoring is decided formally at gates and recorded, because the baseline is the agreement. In adaptive delivery, tailoring is continuous and team-owned: the team adjusts its own practices in review, and the retrospective is the record. In crisis, the cycle compresses: the canvas is one page, the review is daily, and decision rights sit as low as they can go without breaking the floors. Regulated environments invert the emphasis: the obligations are not tailorable, only the manner of meeting them is. What is constant across all four is the record: a tailoring decision that is not recorded cannot be examined, challenged, or learned from.

A language model can draft a tailoring canvas from a practice inventory in minutes, flagging duplicates and suggesting lighter forms. It cannot know which practices are legal, contractual, or safety obligations, and it will be fluent about dropping things; treat its proposals as hypotheses, verify each against the obligations register, and keep the floors in human hands. It can also draft the ethical decision record, which is only honest if the decision actually happened.

The tailoring canvas at Northstar

Daniel runs the evening session as a tree, not a debate. Sanctions screening is required by law and every donor contract, so it stays, run overnight against the official list: a registry query, not a week. Driver and vehicle safety protects people, the drivers and everyone on the flooded route, so it stays in function, lightened in form: licenses verified against the registry, a roadworthiness inspection at the hub before the first convoy. The cold chain stays: two loggers per refrigerated load, a spot-check on arrival. Beneficiary data stays locked, with named access. Contract terms are simplified, not abandoned: no half up front, staged payments tied to verified deliveries, and a first wave of six trucks as a pilot while the remaining thirty-four are verified in parallel. Donor reporting compresses to a one-page exception-and-decision brief three times a week, because the fact of what was spent and where is a floor and the length is not.

The table that comes out of the session is the control-intensity matrix: for each practice, what it protects, its minimum viable form, what is never cut, and the trigger that escalates.

The arithmetic changes the meeting, as it did at BlueLine in chapter 3. The hub is 16 tons a day short, about 160,000 units of supplies a day. Full due diligence costs six days on average: about 960,000 units of unmet need before the first new truck moves. The fast track costs two days: about 320,000. Skipping the checks entirely saves two days over the fast track, about 320,000 units: real, but smaller than it feels. The red flags are not neutral: Farida estimates roughly one chance in eight, about 12 percent, that the firm is not what it claims. If it fails, the exposure is the 900,000 up front, spoiled or stolen first cargo of about 400,000, and four days of rerouting at about 640,000: roughly 1.9 million units, plus an unquantified share of harm to people. One eighth of that is about 240,000 units, close enough to the 320,000 saved that the arithmetic cancels, which is the honest result. When the numbers are close, the decision rests on the floors: sanctions, safety, safeguarding, and medicine integrity, none of which can be priced into an expected value. The fast track gets most of the time benefit with a fraction of the exposure, and it keeps every floor intact. These figures are illustrative and the probability is a judgment call; the structure is the lesson, not the precision.

The decision is made, and Daniel writes the record. The context: the lapsing window, the rationing hub. The options: full due diligence, the fast track, and the skip, which the room discussed and rejected. Who decided: Daniel, with Farida, Priya, Mateo, and Ines named as consulted. The evidence: the red flags and their strength. What was preserved: the four floors. What was simplified: the contract, the vetting, the reporting. The triggers that would reopen the decision: a failed first load, a license check that fails, a second red flag from the registry. Who was told: the whole field team, with an open door to challenge. The record takes fifteen minutes, and it is the reason the decision can survive a review in six months.

When the emergency becomes the habit

The failure this chapter fears most is not the single bad call but the blanket cut: under pressure, a team decides that a whole category of practice, documentation, testing, review, vetting, is bureaucracy, and drops it everywhere, regardless of consequence. The tell is uniformity: the same practice is dropped for the vaccine shipment and the stationery order alike. The blanket cut feels like decisiveness and is actually the refusal to do the one thing tailoring requires: weigh each practice against what it protects.

Its mirror is performative compliance: the form kept with none of the function, the checklist signed without a check, the review held without a decision. The two failures usually coexist: ceremony where nobody cares, nothing where somebody does. Both are solved by the same instrument: run every practice through the tree and record the decision.

There is a third failure that arrives later. The fast track justified for the flood becomes the standing procedure, and no one revisits it: the two-day vetting that protected a crisis becomes the permanent vetting that protects nothing, because the triggers that would reopen the decision never got a date. Every tailoring decision should carry a revisit condition; Daniel’s record carries the trigger list and a review date thirty days out. The habit of reviewing the cuts is what keeps an emergency response from hardening into an excuse.

Practice

1. A triage of practices. For each practice, decide: preserve (nonnegotiable), simplify (keep the function, lighten the form), or cut with a recorded replacement. (a) Sanctions-list screening of every new supplier. (b) A 30-page monthly risk register update that no one reads. (c) A weekly two-hour steering meeting that makes no decisions. (d) Temperature logs on every refrigerated vaccine shipment. (e) A second signature on every purchase order under 100 units. (f) A daily 15-minute field coordination call by radio.

Guidance: (a) preserve; it is a legal and donor-contract floor. (b) cut or shrink to one page of owned risks with named owners and dates; the replacement is a decision, not a document. (c) cut or convert to an exception-and-decision forum: no status round-robins, decisions on the table, an owner and a date for each. (d) preserve the function, lighten the form with electronic loggers and a spot-check; temperature integrity protects patients. (e) simplify: raise the threshold or move to one signature with an audit trail, because nobody checks small-value spending anyway. (f) keep; it is coordination and a safety decision at the point of work. The common error is deciding by annoyance rather than by consequence.

2. Write the record. Draft the ethical decision record, all seven elements, for this situation. Your sponsor asks you to report the project at 80 percent complete because the board needs a green quarter; your evidence says 62 percent, and the gap is a supplier delay that is fully visible in your data. What do you write, and what do you tell the sponsor?

The record should name the pressure (a board signal, a sponsor’s incentive), the options (report 62 with the supplier evidence, report 80, or delay), who decides, the evidence and its strength, what is preserved (no managed metrics), what would change the decision (a credible recovery plan, a re-baseline the board is told about), and who is told, including the supplier. The unsafe answer keeps the number honest and the narrative misleading; the record exists precisely so the two cannot diverge. A defensible alternative: report 62, attach the recovery plan, and give the board a range, 62 to 72, with the confidence stated.

3. The tailoring meeting. Your team proposes to cut documentation, testing, and review to hit a date. Run each cut through the tree, name what each protects, and decide preserve, simplify, or escalate, with an owner and a revisit condition for each. Then say what goes into the decision record.

The defensible pattern: testing that protects safety or acceptance keeps its function and lightens its form, the critical tests run and the exhaustive ones deferred with a trigger; documentation that changes a decision or transfers accountability stays at minimum length; review that produces only signatures becomes a decision log; and anything touching a contract, regulation, or safety obligation is escalated before it is cut. The common failure is the blanket cut of all three together; credit goes to answers that name what each cut protects and record the trigger that would reopen it.

4. The mastery drill. A second surge has flooded the southern road, and you lead the response. Four requests arrive together: the regional government asks for the full shelter registration list “to coordinate aid”; the field lead proposes skipping driver license checks “just once” to get medicine moving; the donor’s office demands a weekly report while your communications are down; a local firm offers thirty trucks at half the market rate with no records of any kind. For each, decide what to simplify, what to preserve, and what to escalate.

Registration data: preserve privacy. Offer the government aggregated shelter counts and needs, and route the raw list through a protection assessment; a named official and a purpose statement are the floor, not theater. Driver checks: preserve the function. The simplified form is a ten-minute registry verification before each convoy, fast but never optional. Reporting: preserve the fact of what was spent and where, simplify the format to one page delivered by whatever channel works, radio, courier, or a hand-carried note, and escalate the broken communications as an operational risk, not a reporting inconvenience. The trucking offer: run it through the tree; with no records, the red flags are the offer itself, so fast-track with staged exposure and a first-load inspection, or decline and say why. Unsafe answers: handing over raw personal data without a protection assessment, one unlicensed convoy, invented report numbers, or a half-value advance to an unverifiable firm. The drill is passed when every floor is named before any speed is claimed.

5. Transfer question. Take the project you lead or work in now, and run the tree on the three practices you are most tempted to cut. Name one piece of ceremony you will retire and what replaces it, one floor you will defend, and a revisit date for each cut.

Tailoring is not the art of removing things. It is the art of deciding what to protect, in a form the situation can afford, with a record that makes the decision examinable. Give every cut a trigger and a date, so the emergency response does not harden into an excuse.

There is an earlier question than how to run the work, and it is whether to run it at all. Chapter 5, “Select the Right Work,” moves upstream: before you tailor a delivery system, you must decide whether the work deserves scarce resources, which is where judgment first pays.

Notes

• Northstar Relief Logistics is a composite case created for this book; it depicts no real organization, region, flood, or firm. All characters are composite, and all figures in this chapter are author-created illustrations for teaching.
• The do-no-harm principle summarizes the case-based research in Mary B. Anderson, Do No Harm: How Aid Can Support Peace or War (Lynne Rienner, 1999).
• The Sphere Handbook: Humanitarian Charter and Minimum Standards in Humanitarian Response (Sphere Association, 2018) is the humanitarian sector’s shared set of minimum standards; readers should consult the current edition for the official text.
• Goodhart’s observation about target-driven measures originates in a 1975 note on monetary policy by the economist Charles Goodhart; the widely quoted formulation “When a measure becomes a target, it ceases to be a good measure” is Marilyn Strathern’s, from “Improving Ratings: Audit in the British University System,” European Review 5, no. 3 (1997): 305-321.
• ISO 21502:2020, the current general ISO guidance on project management, states that its guidance can be tailored to the needs of the organization and the project; this book is independent of ISO.
• PRINCE2, in its current seventh edition (PeopleCert), names tailoring among its principles; readers should consult PeopleCert’s official materials for the current text.
• The November 2020 Scrum Guide, by Ken Schwaber and Jeff Sutherland, describes Scrum’s roles, events, artifacts, and rules as immutable: partial implementation is possible, but the result is not Scrum. Tailoring in Scrum happens around the framework, not inside its rules.
• The PMBOK Guide, Eighth Edition (Project Management Institute, November 2025) describes six core principles and seven performance domains; readers should consult the official publication for the current text, as this book is independent of PMI.