How we thinkArticle

The order is not the decision.

When a ranking will not stay settled, the cause is almost always in what went in, and almost never in the list.

In this piece
  1. A list is what is left when the argument ends
  2. Two things go in: what matters and what is true
  3. The same beliefs give a different order under different values
  4. So a disagreement about the order is one of two disagreements
  5. Authority is a separate question from both
  6. Why arguing the list feels like progress
  7. What changes when the inputs are written down
  8. What this asks of the meeting
  9. Where a ranking stops being enough
  10. Where this argument is weakest
The order is not the decision. It is the arithmetic of the values and beliefs that went in.

There is a meeting most organizations hold at least once a year. A list goes up on a screen. It has more items on it than there is capacity to do. Over two or three hours the items move. One goes up because its sponsor is in the room. One goes down because its sponsor is not. Near the end, when people are tired, the order that happens to be on the screen becomes the plan.

A few weeks later the list is open again. Someone new has joined, or a customer has called, or a team has found that the thing ranked fourth cannot start until the thing ranked ninth is finished. The same items move again, for the same reasons.

It is tempting to read this as a failure of discipline. We think it is something simpler. The meeting argued about the wrong object.

A list is what is left when the argument ends

Every organization has more claims on its people, money, attention and time than it can meet. A priority is what exists when one of those claims has been resolved against the others: this before that. Until something has been placed ahead of something else, for a reason, there is a wish list and there are no priorities.

The list on the screen records the result of that act. It does not record the act. It shows that onboarding is first and the platform work is fourth. It does not show why, and the why is the only part that can be defended later.

Two things go in: what matters and what is true

Behind any order there are two kinds of input.

The first is what the organization values. In practice this takes the form of criteria, and of how much weight each one carries. Is this a year for growth or for stability? Does a dollar of new revenue count for more than a dollar of cost avoided? These are choices. Evidence can inform them. It cannot make them.

The second is what the organization believes. For each item: how much it would help on each criterion, whether it can be done, what it will cost. These take the form of scores and the evidence behind them. They are claims about the world, and they can be right or wrong.

Put together, the two produce an order. That is all a ranking is. The DEGAP method places ranking last of its five phases for this reason. The first four settle who is deciding, who should shape the decision, what is in consideration, and what the criteria are.

The same beliefs give a different order under different values

An invented example makes the point. Four pieces of work compete for one team. Three criteria matter: effect on customers, revenue this year, and risk removed. Each item is scored from 1 to 5 on each.

ItemCustomersRevenueRisk
Rebuild onboarding522
Enterprise billing253
Mobile redesign421
Retire the old platform215

Suppose the leadership has said this is a growth year. Customer effect carries half the weight, revenue 30 percent and risk 20 percent. Later, after a bad outage, the same leadership decides stability comes first. Risk now carries half the weight, revenue 30 percent and customer effect 20 percent. Nobody changes a single score.

RankGrowth yearTotalStability yearTotal
1Rebuild onboarding3.5Enterprise billing3.4
2Enterprise billing3.1Retire the old platform3.2
3Mobile redesign2.8Rebuild onboarding2.6
4Retire the old platform2.3Mobile redesign1.9

The item that was first is now third. The item that was last is now second. No one learned anything new about any of the four. What changed was what the organization decided to care about.

The reverse also happens. Hold the growth-year weights and let one belief change. Finance reports that enterprise billing will bring in less this year than assumed, and its revenue score falls from 5 to 3. Its total drops from 3.1 to 2.5 and it slips from second to third. That is a change in the order with no change in values at all.

So a disagreement about the order is one of two disagreements

When two people disagree about where an item belongs, one of two things is true. They weigh the criteria differently, or they believe different things about the item. From across a table the two look identical. Both sound like “this should be higher.”

They are different arguments, and they belong to different people. A disagreement about weights is a conversation about direction. It belongs with whoever has the standing to set direction, and it ends with a choice. A disagreement about scores is a conversation about evidence. It belongs with whoever knows the work, and it ends when someone goes and finds out.

If we agreed on the facts about this item, would we agree on where it goes?

That question is a useful test. If the answer is yes, the disagreement is about beliefs, and more evidence can settle it. If the answer is no, it is about values, and no amount of evidence will.

Run as one argument, with everyone present, neither gets resolved. The people with evidence are overruled by the people with authority, on questions of fact. The people with authority are worn down by detail, on questions of direction. The meeting ends when the time does.

Authority is a separate question from both

There is a third thing that is easy to confuse with the first two: who decides. Involving many people improves the inputs. People close to the work hold better beliefs about it, and people close to customers know what the criteria should capture. But taking part in shaping a decision is different from having the standing to make it.

When that difference is left vague, a ranking can be well reasoned and still not hold, because nobody who could commit the organization actually did. It helps to settle, before any scoring starts, who will make the final call on weights, who owns each score, and who can reopen the result. Those are short answers. Leaving them unstated is expensive.

Why arguing the list feels like progress

Moving items on a list has real appeal. It is concrete. Everyone can see the change. Each person can speak for the item they know. At the end there is something to show.

The trouble is what it leaves out. The list holds positions and no reasons. So the first time a position is questioned, there is nothing to consult. The only way to answer “why is this fourth?” is to hold the argument again, usually with a slightly different group, who reach a slightly different order. That is what a reopened ranking is: a list being asked a question it cannot answer.

What changes when the inputs are written down

If the weights and scores are on the record, three things become possible.

A change gets an address. “We moved the platform work up” becomes “the weight on risk went from 20 to 50 after the outage.” A person who disagrees knows what to disagree with.

A change can be explained to people who were not in the room. The teams who do the work are rarely the ones who set the order. Given the reason, they can apply it to the hundred smaller choices the list does not cover. Given only the order, they can comply or quietly not.

The next change gets cheaper. Priorities are supposed to move when the facts or the direction move. When the inputs are recorded, a new fact changes one score and the order follows. Nobody has to rebuild the whole list from memory.

What this asks of the meeting

In practice the change is mostly one of sequence.

First, settle who is deciding and what question they are deciding. Then agree on the criteria and how much each one counts. Do this before the items are scored, while nobody yet knows which item a given weight will favor. Then score. People who know the work score it. Where they disagree, they say what evidence would settle the matter and who will get it.

Only then look at the order. If it surprises people, that is useful. A surprise means a weight does not reflect what the group values, or a score does not reflect what someone believes. Go back to that input and change it openly. Do not reach into the list and drag the item.

Dragging the item is the moment the record stops being true. From then on the list says one thing and the stated reasons say another, and the next person to ask why will get an answer that does not add up.

Where a ranking stops being enough

Everything above treats items as separate. Often they are not. One cannot start until another finishes. Two only make sense together. Three need the same four people in the same quarter. These are beliefs too, but they are beliefs about items in combination, and a sorted list cannot hold them. An item can be correctly ranked fourth and still be the first thing that has to start.

At that point the decision has to be made across the whole set at once: which combination of work, in what sequence, fits the capacity there is. The ranking remains an input to that. It is no longer the plan.

Where this argument is weakest

Three cautions are fair.

Numbers can lend a false precision. A total of 3.1 against 2.8 is not a measurement. The scores are judgments, and the arithmetic is a way of keeping those judgments consistent and visible. If small differences in totals are deciding things, the fair reading is that the items are tied.

Weights can be fitted after the fact. Anyone who knows the order they want can find weights that produce it. The protection is sequence: agree on the criteria and weights before anyone sees the ranking they lead to.

And not every decision needs this. A team of six that talks every day can hold its reasons in conversation. The case for writing the inputs down grows with the distance between the people who decide and the people who carry the decision out.

None of this makes prioritization mechanical. Choosing what to value is as hard as it ever was. What the split offers is a way to have that hard conversation once, with the right people, and to keep what it produced. The order will change. When it does, it helps to know which of the two inputs moved.