🎯 Quick Answer

What: To prioritize improvement projects is to decide which few will get real resource this cycle and to state plainly that the rest will not. Ordering a list is only the first half of that. The second half is declining what falls below the line.

Why it matters: Improvement work has a threshold. A change that receives a fraction of the effort it needs does not deliver a fraction of the benefit. It delivers nothing, and it consumes the fraction on the way. A portfolio of half-worked problems looks like a program and produces almost no output.

How to apply: Treat capacity as the fixed input and the list as the variable. Rank on counted evidence rather than conviction, usually with a Pareto chart. Draw a line. Fund what sits above it properly, name what sits below it as not happening this cycle, tell the affected sponsors directly, and set the date the line gets revisited.

The payoff: Fewer things finish, and they stay finished. Teams that stop the majority of their active efforts routinely clear the survivors faster than they were clearing the whole portfolio.

Every improvement list I have seen is honest. Almost none of them are decisions.

A team does the work properly. They gather the problems, rank them, argue about the order, and arrive at a list everyone in the room can defend. Then the list goes into a plan, and somewhere between the ranking and the plan a quiet thing happens: everything on it gets a workstream. The top item gets one. So does the fourth. So does the ninth, because somebody in the room owns it and leaving it out would be awkward.

That is not a priority list. It is an inventory with an order printed on it.

Ranking is not prioritizing

The two get treated as the same act, and they are not even close.

Ranking is analytical. You compare items, you order them, and if you have done it well the order holds up to scrutiny. It is genuinely useful work and it is also completely safe, because nothing has been taken from anyone. Every problem is still on the board. Every sponsor can still point at their item and see it there, third from the top or eleventh, but present.

Prioritizing is the part that costs something. It is the act of declining the rest, out loud, with a name attached to the decision. It requires telling a colleague that the thing they care about is not this quarter's work. That is why so many teams do the first half beautifully and never do the second: ranking is a spreadsheet exercise and prioritizing is a conversation nobody enjoys having.

So the list grows and capacity does not, and the gap between them gets absorbed by spreading everyone thinner.

The arithmetic nobody runs

Here is the part that makes this expensive rather than merely untidy.

Money divides cleanly. Give five efforts a fifth of a budget each and you have five efforts running on a fifth of a budget. Attention does not behave that way. Improvement work has a threshold, a level of sustained effort below which a change simply does not hold. Above the threshold, a new way of working survives the week somebody is out sick, survives the first busy period, and eventually becomes how things are done. Below it, the change reverts, and everything invested in it leaves with it.

Which means a team running five efforts at a fifth of the resource each does not get five improvements at a fifth of the speed. It gets nothing for months, and then gets whichever one somebody eventually protected long enough to finish.

I have watched this from the inside more than once, and the thing that makes it so hard to spot is that nothing looks wrong while it is happening. Every project has a sponsor. Every project reports progress. Everybody is genuinely, visibly busy. Then the quarter closes, the results slide is thin, and the honest explanation is not that anyone underperformed. It is that a portfolio was never narrowed.

Half-funded is unfunded. It just takes a quarter longer to find out.

What the line does

The mechanism that fixes this is unglamorous. Once the ranking exists, draw a line across it.

Above the line: the few items that will get real resource, real ownership, and enough of both to actually finish. Below the line: everything else, named openly as work that is not happening this cycle. Not "later." Later is the word teams use when they want to avoid the conversation, and an item marked later behaves exactly like an item still on the list. It absorbs a little attention every month, it makes the board look longer than it is, and nobody ever quite kills it.

Three things make the line hold.

Fund above the line generously. More than the items appear to need. An improvement running on a thin allocation stalls quietly, and a stalled improvement returns nothing for what it already consumed.

Tell the sponsors below the line directly. If nobody has heard the word no, the decision has not actually been made. It is a draft, and it will be quietly relitigated by week three.

Set the date the line gets revisited. A fixed review date turns the line into a decision. Revisiting it whenever somebody lobbies turns it back into a queue.

Where the evidence comes from

None of the above works if the ranking itself is a matter of opinion, because a line drawn across an opinion will not survive its first challenge.

This is where a Pareto chart earns its place, and it is why prioritization sits in the Analyze phase rather than in a planning meeting. You count occurrences over a defined window, sort the causes into descending bars, and lay a cumulative line across the top. Where that line crosses eighty percent, the chart has named the small group of causes carrying most of the effect. Everything to the right of the crossing is not being ignored. It is being deferred, on counted evidence, in a form a sponsor can read in ten seconds.

That is the difference between a defensible line and an arbitrary one. When somebody asks why their item is below it, the answer is a number rather than a preference.

Two cautions are worth carrying. A Pareto is only as good as the counting, so every category has to be defined tightly enough that two people classifying the same event agree; if raters disagree, the ranking is noise wearing a chart. And if every bar comes out roughly the same height, there is no concentrated group to attack. That is a real finding rather than a failed analysis. It says the problem is spread, and a broad systemic fix will beat a targeted one.

It also assumes the work upstream was done honestly, which means you listed every plausible cause before you counted them, and that the items you are ranking are causes rather than symptoms. Ranking symptoms produces a confident order and no result, for the same reason that fixing one does: you can only get to a root cause that somebody named.

The project nobody ever stopped

There is one item on almost every board worth handling separately, because it will not be caught by any ranking exercise.

It has been there longer than anyone remembers starting it. It gets carried forward every planning cycle. It has an owner, technically. It has never had enough resource to move and has never been formally stopped, so it survives by default, year after year, absorbing a small amount of attention every month and a slightly larger amount of guilt.

Nobody keeps it alive on purpose. Stopping something requires a decision and a conversation. Carrying it forward requires neither. And the real cost is not the work it consumes, which is small. The cost is that its presence makes the whole list feel immovable, and that feeling is exactly what stops a team from cutting anything at all.

Take it down. Say plainly that it is not happening. If it matters again in six months it will come back on its own merits, and it will come back with resource, which it never had.

What this looks like when it works

The pattern is consistent enough that I now expect it. A team stops the majority of what it has running, gives everything to the two or three that matter most, and closes them faster than anyone predicted. Then, over the following period, it works through the remainder at several times the rate it was managing while running all of them at once.

The projects that got stopped were rarely bad projects. That is the part people find hard. Most of them were good, and stopping them still made the program work, because the constraint was never the quality of the ideas. It was that nothing had been chosen.

If you want to prioritize improvement projects in a way that survives the quarter, the question to put to the room is not which of these matters most. Everyone already knows that, and answering it changes nothing. The question is: what are we choosing not to fix this cycle, and who needs to hear that from us directly?

I write more field notes like this, and share the tools I build, in my newsletter. You can sign up here.

Frequently asked questions

What does it mean to prioritize improvement projects?

It means deciding which few efforts will receive enough resource to finish this cycle, and stating explicitly that the rest will not. Ordering a list by size or cost is only the analytical half. Prioritization is complete when something has been declined and the affected sponsors have been told.

How many improvement projects should a team run at once?

Fewer than feels comfortable. The useful test is not a fixed number but the threshold: can each active effort be resourced to the level where the change would survive a busy period and a key person's absence? If not, the portfolio is too wide, whatever the count.

How do you decide which improvement projects to stop?

Rank on counted evidence rather than conviction, usually with a Pareto chart over a defined window, then draw a line based on capacity rather than ambition. Ask of each item below the line what it would take to close it properly this month, and stop anything you are not prepared to fund at that level.

Isn't stopping a project a waste of the work already done?

It is the opposite. Work already spent is spent either way. Continuing to spread resource across an effort that cannot reach its threshold adds cost without adding benefit, so the choice is not between finishing and stopping. It is between stopping now and stopping later after paying more.

Where does the Pareto principle fit into prioritization?

It supplies the evidence that makes a decision defensible. When a small number of causes carry most of an effect, choosing those causes is not neglect of the rest, it is the shortest route to the result. It also gives you something concrete to show a sponsor whose item fell below the line.