Outcomes or delivery health? The metrics fight inside every PMO, argued both ways

The quarterly metrics review starts on time, which feels like a good omen. The PMO lead shares her screen. Fourteen initiatives, thirteen green. Cycle time trending down for the third straight quarter. Say-do ratio in the nineties. Escape rate flat. It's the healthiest portfolio slide that room has seen all year, and she's earned it.

The sponsor lets her finish. Then he asks the question the dashboard doesn't have a cell for.

"Then why does it feel like nothing landed this quarter?"

Renewals didn't move. The cost line didn't move. The launch everyone toasted in January is live, technically, and the customers it was built for haven't noticed. The room goes quiet in that specific way rooms go quiet when everyone knows the next sentence matters and nobody wants to own it.

Here's what almost nobody says out loud in that meeting: both of them are right. The delivery numbers are real. The flat quarter is real too. The dashboard's version of the quarter and the sponsor's version of the quarter are both true, and they never touch each other.

Out of that meeting comes the oldest fight in the PMO. One camp says measure outcomes, full stop. Green dashboards that produce nothing are exactly the disease, so stop measuring the machinery and start measuring the money. The other camp says measure delivery health, because outcome numbers arrive too late and explain too little to run a portfolio with. I've sat in both camps at different points, which is embarrassing to admit and useful for what follows. So let's stage this properly. Each side gets its strongest case, no strawmen. Then I'll rule.

The outcomes corner: nobody funds you to move cycle time

Start with the purest question in the building: why does this initiative exist?

Not to burn down story points. Not to improve flow. It exists because somebody believed the company would make money, save money, or stop losing money by funding it. That's the entire arrangement. Outcomes aren't one lens on the work. They're the reason the work is allowed to cost money at all.

Measure anything else as your headline number and you're farming theater. Anyone who's run delivery long enough has cut into a watermelon dashboard: green on the outside, red the second you slice it. Statuses are green because humans report them, and humans report whatever makes the meeting end sooner. And before the health camp gets smug, its metrics rot exactly the same way. The quarter leadership starts watching cycle time, tickets get split smaller. The quarter escape rate goes up on the big screen, bugs get rebadged as enhancements. Goodhart's law isn't a risk you manage. It's a delivery date: every proxy metric will eventually be gamed, and the only open question is which quarter it happens in.

Outcomes resist that rot in a way proxies never will. You can reclassify a bug. You can't reclassify flat revenue. The renewal number does not care about your definition of done.

And beneath all of that sits the coldest argument, the one that ends budget meetings. Executives fund outcomes. Sooner or later the portfolio gets cut, and when it does, the initiative that can name the business number it moved survives, while the one with the beautiful flow charts gets consolidated into it. Try defending a budget with a cumulative flow diagram and watch the CFO's face.

A team that hits every delivery metric and moves no business number isn't healthy. It's punctual.

That's the outcomes case, and notice what it never claimed. It never said the engine doesn't matter. It said the engine can't be the headline, because headlines corrupt whatever stands under them.

The health corner: you can't steer with the rear-view mirror

Now the other corner, and it swings just as hard.

Outcomes lag. Not by a sprint. By quarters. The work your teams shipped in March surfaces in a renewal conversation in September, filtered through a sales cycle, a pricing change, and somebody else's fiscal year. Manage a portfolio on outcomes alone and you're steering with the rear-view mirror: perfectly accurate about where you've been, useless about what to do next.

Worse, short-term outcome numbers lie in both directions at once. A good quarter can hide a broken engine, because the revenue landing today was won by decisions made two years ago while cycle time quietly triples underneath it. A bad quarter can punish good work, because the team shipped exactly the right thing into a market that moved sideways. Reward on outcomes alone and, over enough quarters, you'll bonus the lucky and punish the sound. That isn't accountability. That's a casino with OKRs on the wall.

Then there's attribution, the problem the outcomes camp keeps stepping around politely. When a business number moves, who moved it? Marketing repriced. Sales restructured territories. Three product teams shipped in the same quarter. A competitor stumbled over their own migration. Pinning that number on one initiative is mostly fiction with a spreadsheet attached. The outcomes corner just finished telling you that proxies get gamed. Credit gets gamed too, and usually by more senior people.

Meanwhile, the metrics the outcomes camp just sneered at hold the one property an outcome will never have: somebody can act on them on Tuesday morning. Cycle time doubled in two sprints? Go look, because something specific broke: an approval step, a dependency, one reviewer buried alive. Predictability sliding? Planning and reality have stopped speaking to each other. Escape rate climbing? You're shipping faster than you're testing. Every one of those is a lever a named human can pull this week. "Revenue is flat" is not a lever. It's a verdict, delivered months after the decisions that earned it.

Health metrics are the gauges on the engine. Nobody flies a plane by staring at the destination.

The ruling: it was never a real choice

Here's the truth about this whole fight: each camp is arguing against the other side's failure mode, not against its actual position.

The outcomes camp is right that health metrics with nothing behind them curdle into theater. The health camp is right that outcomes with nothing underneath them are laggy, unactionable, and unfair. Neither camp has shown you have to choose, because you don't. Nobody in the outcomes corner genuinely wants a portfolio with no gauges. Nobody in the health corner believes flow is the purpose of the company. The fight is real. The binary is not.

So here's the ruling, and it fits on an index card.

Every initiative carries one outcome it serves. One. Not a wall of OKRs. One business number this work exists to move, named plainly enough that a new joiner could repeat it in the elevator.

Next to it, two or three health indicators chosen for that team's engine: cycle time, predictability, escape rate, flow efficiency, whichever gauges actually describe how this particular engine runs. Enough to see the engine. Few enough that nobody can hide inside them.

Both go on the same page and get read in the same meeting. Outcome first, then health, initiative by initiative, with one question stapled to the review: do these two stories agree? The moment outcomes report to the steering committee while health reports to the delivery sync, you haven't settled the fight. You've given each camp its own room and better stationery.

And never bonus either side alone. Tie money to outcomes alone and scope gets gamed: targets get sandbagged, tides get claimed as swimming, and every initiative drifts toward whatever counts instead of whatever matters. Tie money to health alone and classification gets gamed: bugs become enhancements, tickets shrink to look faster, and the watermelon comes back wearing a lab coat. A bonus wired to a single metric is Goodhart's law with a payout date. Use the pair to inform judgment, and keep it off the payroll.

None of that is a compromise between the camps. It's the instrument each of them was describing half of.

The whole point is the disagreement

Because here's what the pair buys you that neither number buys alone: a forced, honest conversation on the day the two stories split.

Read the combinations like a board.

Health green, outcome moving. Carry on, and stay slightly humble about attribution, because some of that number is tide, not swimming.

Health red, outcome flat. Painful, but coherent. The engine is struggling and the results show it. Fix the engine before anyone touches the strategy.

Health green, outcome flat. The most dangerous cell on the board, because it photographs like success. Everything ships and nothing lands. You're building the wrong thing well, and that isn't a delivery problem. It's a bet problem, and it belongs to whoever chose the outcome, not to the team hitting its marks.

Health red, outcome moving. You're borrowing from next quarter. The number landed while the engine degraded: old momentum, heroics, a market being temporarily kind. Take the win, then fix the engine before the loan gets called.

Now go back to the meeting this post opened with. Green board, furious sponsor. That's the third cell arriving as an ambush. The metrics didn't fail that portfolio. The two halves of the truth were never asked to sit in the same room, until a sponsor forced them to, eight months late and in front of everyone.

So take the fight in on purpose. One outcome per initiative. Two or three gauges under it. One meeting where both get read aloud, and no bonus riding on either alone. When the numbers agree, move faster. When they disagree, the disagreement is the agenda, not a reporting error to reconcile quietly before anyone important sees it.

A PMO that measures only outcomes can't steer. A PMO that measures only health can't matter. The job was never to keep the dashboard green. The job is to make sure the sponsor's question never arrives as a surprise.