Day 66 · 2026.07.25

Public & Nonprofit Leadership: Leading Where There Is No P&L

Topic: Public & Nonprofit Leadership·4 principles
"In the social sectors, money is only an input, and not a measure of greatness." — Jim Collins
Why this lands on your desk: a large slice of any big company runs on nonprofit logic — platform teams, internal tooling, open source, cross-org initiatives, ERGs. They share the three defining traits of a nonprofit: no revenue line to serve as a scorecard, part of the workforce is volunteer (20% time, not yours to review), and accountability to several people who don't sign your paycheck. The failure modes are identical too: reporting a pile of activity and still losing budget, trading mission for unpaid overtime, volunteers evaporating within two weeks, and promising four parties four things and defaulting on all of them. Four problems, four tools.
PRINCIPLE 01

No Profit Line: Report Activity, Output and Outcome as Three Separate Layers Inputs Are Not Outputs

metricslogic modelreporting
In a business team, money is both the input and the scorecard. In a nonprofit structure, money is only an input. You have to build your own ruler — admit up front that it is a proxy, then commit to not swapping it for at least three quarters.
"In business, money is both an input (a resource for achieving greatness) and an output (a measure of greatness). In the social sectors, money is only an input, and not a measure of greatness." — Jim Collins, Good to Great and the Social Sectors (2005), p.5
Setting: quarterly review. Your VP asks, "What's the ROI on the platform team?"
✗ The common answer

"We support 40 business teams, we closed 2,000 tickets last year, availability was 99.95%." — all activity. The VP still has no idea what happens if you disappear, so at headcount time you get discounted as a pure cost center.

✓ The rewrite

"Let me give you three layers. Activity: 2,000 tickets — I don't count that as an achievement; the higher it goes, the worse our self-serve docs probably are. Output: median deploy time across those 40 teams went from 22 minutes to 6. Outcome: those 40 teams doubled release frequency this year and cut rollback rate 30% — roughly 9 engineer-years recovered."

"Now the counterfactual: if the platform team froze for a quarter, the first thing to break isn't an outage — it's that curve sliding back, visible in about six weeks."

"And the self-incrimination: this ruler can be gamed — deploy time can be bought by cutting security checks. So I report security-check coverage alongside it. Please read the two numbers together."

The Logic Model
INPUTSpeople · budgetnot an achievement
ACTIVITYtickets · meetingsmost misreported
OUTPUTdeploy time · coverageyou control it
OUTCOMErelease frequencyreport here
IMPACTbusiness growthdon't claim it alone
Report only at "activity" and you are defenseless at budget time; claim "impact" and the business lines will call it credit-stealing. Hold the output and outcome layers.
  • Can I answer, in one number, "if this team froze for a quarter, what is the first measurable bad thing and how soon does it show up"?
  • What was the highest layer in my last report — activity, output, or outcome?
  • What is the cheapest way to game my ruler? Did I volunteer the counterbalancing metric myself?
  • Is there one internal customer willing to restate this outcome in their own words?
  • Treating "not precisely measurable" as "not worth measuring." Saying "our value can't be quantified" is voluntarily surrendering your claim on budget.
  • Passing off satisfaction surveys as outcomes. Internal NPS is a feeling, not a result — it works only as an early warning.

Do: write one counterfactual sentence and send it to your boss: "If this team froze for a quarter, the first measurable failure would be ___, visible in about ___ weeks."

Ask: if you can't write that sentence, is the gap in your metrics — or in how the team is scoped?

PRINCIPLE 02

The Double Edge of Mission: Mission Cannot Buy Overtime Passion Is Not Consent

missionmotivationethics
A mission can substitute for part of the pay. That is both its power and its moral hazard. The better you are at articulating mission, the more it is your own job to police the exploitation line — because your team won't stop you.
"Understanding contemporary forms of exploitation: Attributions of passion serve to legitimize the poor treatment of workers." The experiments found that once someone is seen as passionate about their work, observers are markedly more willing to consider unpaid overtime and out-of-scope tasks reasonable for that person. — Kim, Campbell, Shepherd & Kay, Journal of Personality and Social Psychology 118(1), 2020
Setting: two weeks before a platform milestone. The team has already been overloaded for two sprints.
✗ The common line

"I know everyone is exhausted, but we're building the foundation for the company's next two years — chances like this are rare. Two more weeks of pushing." — this does two bad things at once: it trades mission for unpaid labor, and it silently converts anyone who declines into someone insufficiently committed.

✓ The rewrite

"I'm asking for extra time over the next two weeks. I'll say the mission part once: this is the foundation for the next two years. Everything after this is consideration, and all of it is verifiable —"

"(1) I already got X and Y pushed to next quarter with the VP; notes are here. (2) The week after launch, nobody is on call and no meetings are scheduled. (3) Today I'll draft the exact sentences about each person's contribution that go into your perf packet, and send them to you to check the wording."

"(4) If you have something personal in these two weeks, just say 'I can't' — no reason required, and it affects no evaluation." — that last line is the whole point. Without a cost-free exit, the first three are just packaging.

  • Does the extra effort I'm asking for have verifiable consideration attached (scope cut / time back / written into perf)?
  • Did I give people a cost-free way to decline? Will the ones who decline be quietly marked?
  • Of the promises I made at the last rally, what fraction did I keep? Below 100%, pay the debt before rallying again.
  • If this intensity ran for 12 months, who leaves first? Can I live with that name?
  • Treating "he's passionate" as a reason to give less. This is exactly the bias the study measured, and the person imposing it usually can't see it.
  • Martyrdom culture. Scoring loyalty by hours burned drives out your highest-market-value people first — they have other options.

Do: dig out the promises from your last rally, mark each kept / unkept, and make the unkept one your top priority this week.

Ask: if your team were replaced tomorrow by contractors who honor contracts and not missions, which of your current practices would collapse instantly? Those are the management debts mission has been subsidizing.

PRINCIPLE 03

Volunteers Are Not Cheap Employees: Manage the On-Ramp and the Credit, Not the Tasks Volunteers Vote With Their Feet

volunteersopen sourcecross-org work
You cannot performance-manage a volunteer. You can only lower the cost of their first contribution and guarantee their credit. With volunteers, your center of gravity moves from assigning tasks and chasing status to building on-ramps and handing out recognition.
"Treating your users as co-developers is your least-hassle route to rapid code improvement and effective debugging." — Eric S. Raymond, The Cathedral and the Bazaar, Lesson 7
Setting: you kick off a cross-org initiative (unified logging standards). Five teams each send one person at 20% time. Two weeks later exactly one is still moving.
✗ The common setup

A 30-person channel, a weekly one-hour sync, and a "who owns what" spreadsheet — then you wait for homework. For a volunteer, the meeting is pure cost and the spreadsheet is pure burden; their perf is written by their own manager, and your spreadsheet has no force.

✓ The rewrite (three moves)

1. Build a 15-minute on-ramp: "I split the first 12 services to migrate into 12 independent PRs, each with a template and a single command — about 15 minutes of work. Links are here, and you can contribute without attending anything."

2. Outsource the credit to their manager (the only real currency you hold — and promise it at kickoff): "Every month I send each contributor's manager a three-sentence email spelling out what they did and how many teams it affected, cc'ing them."

3. Give them something with their name on it: the Authors line in the doc, the changelog acknowledgment, co-presenter slot at the internal tech talk.

And convert the sync from "60 minutes weekly" to "25 minutes every two weeks, with every decision written into the notes." Nobody falls behind by not attending, so the meeting stops filtering out your volunteers.

The Contributor Funnel · where you lose them, and your lever
Hears about it
nobody forwarded it → announce it in their manager's staff meeting
Tries once
the env won't build → a first task that finishes in 15 minutes
Comes back
PR sat three days unreviewed → a 24-hour response commitment
Becomes an owner
the credit went to you → hand over the module and the name together
Volunteers almost never drop out at the "willingness" layer. They drop out at friction and at neglect. Fix review latency before you recruit anyone else.
  • How many steps from "hears about it" to "submits a first change"? More than three, cut some.
  • Did every contributor's manager get concrete feedback from me this month?
  • How long does a contributor's PR wait for its first response? Over 24 hours, fix that before recruiting.
  • Is attending my meeting a precondition for contributing? If so, the design is wrong.
  • Managing 20% people like headcount. Put them in a sprint, chase their burndown, and they are gone next week.
  • Putting volunteers on the critical path. The moment an SLA or a backward-compatibility promise appears, it needs funded people.
Female Leader's Note Babcock, Recalde, Vesterlund & Weingart (American Economic Review, 2017) found that low-promotability tasks are asked of women more often, and women accept them more often — both mechanisms operate: a higher probability of being asked, and a higher social cost of saying no. Cross-org initiatives, coordination, note-taking, interview loops and onboarding are textbook examples of that category. Two moves:
(1) As the organizer: stop defaulting the notes and the coordination to the same people. Write rotation into the charter — and put yourself in the rotation.
(2) As the person being asked: before accepting, ask "how will this show up in my perf packet?" That converts invisible labor into a verifiable commitment; and an answer you can't get is itself a reason to decline.

Do: pick one cross-org initiative, build a single first task that finishes in 15 minutes, and send the first three-sentence email to each contributor's manager.

Ask: if every volunteer withdrew tomorrow, where would this work stop? That stopping point is exactly what you should be asking funded headcount for.

PRINCIPLE 04

Five Bosses: Shifting From Executive to Legislative Leadership Legislative Leadership

accountabilitystakeholderspower
When nobody — including your nominal boss — holds enough structural power to decide for you, you have to change gears: not authority, but persuasion, shared interest and political credit. This isn't weakness. It's the physics of the position.
"In executive leadership, the individual leader has enough concentrated power to simply make the right decisions. In legislative leadership, on the other hand, no individual leader—not even the nominal chief executive—has enough structural power to make the most important decisions by himself or herself." — Jim Collins, Good to Great and the Social Sectors (2005), p.11
Setting: the platform team has six engineer-quarters of capacity. The VP wants cost down. Business lines A and B each want a mutually exclusive feature. Security wants three compliance gates. All four asks total eleven engineer-quarters.
✗ The common approach

Reassure each one privately: tell the VP you'll cut cost, tell A they'll get it in Q3, tell B you'll try for Q3 too, tell Security you'll do whatever they say. Three weeks later all four believe you lied to them — and you never told a single lie. You just never let them see each other.

✓ The rewrite (one memo, not four side conversations)

"Capacity this quarter is six engineer-quarters. The four asks total eleven. My proposed ordering, with reasons: security gates (3) — a hard constraint, and not doing them blocks both A and B; A's X (2) — it also lowers B's future integration cost; cost reduction (1); B's Y moves to Q4, with a temporary workaround in the interim (link)."

"If you think the ordering is wrong, reply by Thursday naming the item you are willing to give up. I will not act on replies that only add."

That last line is the mechanism: it forces every "add" request into a "trade" request. Without authority it is the only boundary you can actually hold — and what it protects is capacity, not your ego.

  • What do they actually want — not what they say they want, but what they are measured on at year end?
  • What can they give me (budget, headcount, cover, endorsement)?
  • Do they hold a real veto, or only an opinion? Am I treating those two groups identically?
  • Is my commitment to them written in the same document as my commitments to the other three?
  • Whose problem did I absorb last quarter? Political credit is paid before it is drawn.
  • Reading "everyone is unhappy" as proof of fairness. It may just mean nobody was persuaded. The evidence of fairness is that they can restate your reasoning.
  • Assuming a public trade-off will offend people. What offends people is a private promise later broken. A public trade-off costs you this quarter; a broken private promise costs you the principal of your credibility.

Do: copy this quarter's commitments to every stakeholder into one page, mark the two that contradict each other, and make one of those trade-offs public to both parties this week.

Ask: does long-run legislative leadership erode your decisiveness? When was the last time your team wanted you to just decide, and you went and gathered input instead?

Deeper Questions

Proxy metrics distort once they enter evaluation (Goodhart). With no profit as final arbiter, how often should the ruler change?
Manage "stable" and "valid" separately. Keep the externally committed metric stable for at least three quarters — swapping it is itself an admission the last one was useless. But publish a gaming audit every quarter: what is the cheapest way to juice this number right now, and did we do it? Only two legitimate reasons to change rulers: the team's mandate changed, or the metric is demonstrably cheap to manipulate. When you do change, publish old and new in parallel for one quarter and let people see the difference themselves.
Can the line between mission and exploitation only be drawn by the exploited? If the team itself is fired up and asks to work late, should the leader stop them?
Yes — not out of paternalism but because of information asymmetry. You know the project's real priority in the VP's head and the actual weight overtime carries at calibration; the team doesn't. So the duty to brake is yours. Two workable lines: sustainability (could this intensity run 12 months? If not, it's borrowing, and borrowing needs a repayment date) and reversibility (is the cost recoverable fatigue, or irreversible damage to health and family?). Passion gets to decide whether to sprint, not how long.
Can the volunteer model (20% contribution) scale? When must it convert to funded headcount?
Volunteers are good for exploration and diffusion, bad for on-call and backward-compatibility promises. Three signals to formalize: an SLA appears; maintenance exceeds about a third of the work; bus factor drops to one. The simplest test is a single sentence: "if it breaks, will someone be woken at 3am?" If yes, it needs funded people. If no, the volunteer model can run for a long time. Put that sentence in the first line of your headcount memo.
Does writing low-promotability tasks into a charter rotation just move the unfairness from gender to seniority?
It does — if the rotation only spans junior levels. Two constraints: the rotation must cross levels and include the organizer; and more fundamentally, make coordination a named, weighted dimension of performance. Rotation only spreads the loss; weight is what removes it. If your organization won't assign weight, tell the team honestly: this doesn't score here, and I'll compensate you with visibility or choice of project — rather than pretending it will be seen.