Last month I sat with a team looking at a full year of work they had no chance of finishing. Every project was on the board. Every project had a champion. And every project felt essential to the person who put it there. They knew they were overloaded — they said so, out loud, more than once. What they didn't know was how to choose. So they weren't choosing. They were carrying all of it and falling behind on all of it at the same time.

This was not a competence problem. These were sharp people, good at the actual work. What they were up against is that prioritization is a negotiation — a room full of people with real stakes, fighting for their own projects, forming and abandoning alliances inside the span of a single meeting. That part never goes away, and no framework dissolves it. Anyone who tells you prioritizing is a clean, rational sort has never had to do it with other people in the room. But the fight turns unwinnable when two very different questions stay knotted together, and nobody had ever pulled them apart for this team.

One question is about the work: does this belong on the list at all? The other is about the people: how is each person going to show up while we decide? Knot them together — argue the work and the power in the same breath — and the meeting runs three hours and settles nothing, because every question about a project is secretly a question about whose project it is. Pulling the two apart doesn't make the politics disappear. It gives the negotiation a floor, so that volume and rank don't get to be the criteria by default.

So start with the work, and start narrower than feels natural. Before you rank a single thing, ask what's actually necessary. Not important. Not valuable. Necessary — the thing that sinks everything else if it doesn't happen.

Think about renovating a house. The roof is not on the priority list. Nobody votes on the roof. But if it's failing, it gets fixed, because every other decision is worthless under a roof that leaks. Necessary work is the roof. It doesn't compete for a slot; it's the reason there are slots at all. Once you've named it, it comes off the ranking entirely — it's in, full stop, and you spend no more energy arguing about it.

But "necessary" is a high bar, and clearing it takes two separate tests. First: is it genuinely required? Not important, not overdue, not something a senior person is attached to — required, meaning something real breaks without it. The honest version of the test is to ask what actually fails if this doesn't happen. If the answer is "nothing fails, we just lose something we wanted," it isn't required. It's a discretionary project wearing a necessity costume.

Second — and this is the test teams skip — can you afford it? People get the direction backwards here. For a necessary item, affordability isn't whether it fits in the capacity you have left after everything else. Necessary work comes first; the discretionary work is what flexes. If it's genuinely required, you don't leave it off because the plate is full — you find a way to afford it, out of money, time, hours, or attention, and you find it by cutting discretionary work until the necessary fits. The roof is failing and cash is tight? You don't nod along and redecorate the kitchen anyway. You cancel the kitchen, and the guest bath, and whatever else it takes, until the roof is covered. That's what necessary means: you make room for it.

Which is why "we can't afford it" is usually a dodge — said honestly, you almost always can, you've just decided not to cut the discretionary work to pay for it. Necessary means it passes both tests, not one or the other. Skip the required question and you'll pour capacity into the cheap and inessential while the thing holding the place up goes unbuilt. Skip the affordability question and you'll pile up "necessary" work without ever confronting what it costs you to actually do it.

Which is why naming the tests is the easy half. Enforcing them is the job. "Necessary" is the most comfortable place in the room to park something you don't want to defend, because the label does the defending for you — it skips the line and the scrutiny both. So the discipline is refusing to let anything through the gate that can't pass both tests, especially when the thing failing them belongs to someone powerful. The moment "necessary" quietly starts meaning "I really want this," the gate is broken, and you're carrying everything again.

Everything that isn't necessary is discretionary, and discretionary work has to be ranked against something. Not vibes. Not volume. Not whoever's most senior. Something the team agrees on out loud, in advance. The team I sat with landed on four: customer impact, business growth, industry influence, and whether it addressed a real employee need. Then they put those four in order — because criteria you refuse to rank are just a word cloud.

What happened next is the part I keep thinking about. Once the criteria were real, the most senior person in the room started cutting his own projects. Nobody pushed him. The criteria didn't care that the projects were his, and he could see they didn't clear the bar he'd just agreed to. That's what it looks like when agreed criteria outrank rank — not power vanishing from the room, but power boxed in by something it signed up for out loud. It's the entire reason you name the criteria before you touch the list: so the decision belongs to the bar, not to whoever's hardest to say no to.

But criteria only hold if the room agrees to fight fair, and fighting fair comes down to two rules. Call them the rules of combat — because the combat is real, and pretending it isn't is exactly how the most powerful person in the room wins by default. Agree to them before you start, or the exercise becomes theater with a spreadsheet.

The first rule: no one gets to decide who isn't also accountable. If you've ever mapped roles with something like RACI or DARCI, this is the D and the A — the decider and the accountable — and the rule is that you keep them attached. The person making the call has to be on the hook for how it turns out. Otherwise you get the worst pattern in any overloaded team: someone with authority and no exposure reaches in, insists a project lives or dies, and walks away before the consequences land. Decision rights without accountability is just weight the team absorbs on behalf of someone who's carrying none of it. No D without the A.

The second rule is about how each person shows up, and it starts with an honest answer to one question: how much do you actually care about this call? There are only three honest answers. One — this matters to me, and I want to be an active decider; I'm in the room, I'll do the work of deciding, I'll own the result. Two — this matters some; I'll weigh in, and then I'll commit to whatever the group decides and not relitigate it in the hallway afterward. Three — I don't care; do whatever you want, and I won't second-guess it later. All three are fine. What isn't fine is refusing to pick one.

Because that's where the failure modes live — with the people who won't declare a posture. There's the silent one, who won't say they care and won't say they don't, and then resurfaces after the decision to explain why it was wrong. And there's the performer, who acts like an active decider — lots of noise, lots of presence — with no intention of owning anything, because what they actually wanted was to be heard, not to be responsible. Both corrupt the signal. The room reads engagement it can't trust and defers to a voice that was never going to stand behind the call.

Do all of this and a team can finally move — not because the work got smaller, but because they have a way to fight the fight fairly instead of by volume, and to keep the choosing honest once the shouting stops.

One last thing, because it's the harder problem and it deserves its own conversation. The list you finally tame doesn't stay tamed. Finished work doesn't leave the board — it converts into something you now have to keep running, and maintenance strolls straight through the necessary gate without anyone asking it to justify itself. So the pile you fought to shrink starts refilling, quietly, with things nobody actually chose. The fight you can see is the winnable one. The trap is everything that gets to skip the fight entirely — and how much of it you put there yourself.