In 1994, nine employees walked into my office and asked me to fire one of their teammates. They told me I would not need to backfill the role. They were right. The structural pattern they showed me is the operating logic of the Kill List.
My first HR job was at Bently Nevada in 1994. Manufacturing and engineering company, about a thousand employees, vibration monitoring for industrial machinery. I was twenty-two, fresh out of UNR, brand-new to the field.
A group of nine employees came into my office one afternoon. They were upset about a teammate. She was always late or absent and had hygiene issues. She was on her third final warning and the supervisor still had not fired her. The nine told me it was not fair. They had carried her for months. They asked me to follow through on the third final warning.
Then they said something I have thought about for thirty years. They said that if we fired her, we would not need to backfill the position. They would absorb the work and increase their individual output within the same hours they already worked. They told me they would be so glad she was gone that the productivity would more than cover the gap.
We fired her. They absorbed the work. Output went up. We did not backfill. The team ran lighter and faster with nine instead of ten. The math worked exactly as the nine said it would.
Boom.
I did not have the vocabulary for it at twenty-two, but the nine employees demonstrated something structural. Output is not just a function of effort. Output is a function of distribution. Removing one drag-source moved the entire distribution upward, faster than adding a tenth person ever would have.
Discretionary effort is not a motivational story. It is a structural one. The nine were not lazy before. They were not motivated after. They were freed. The work that was draining the team was draining them in a specific, named way. Remove the drain. The output returns to where it would have been if the drag had never been there.
The cost of carrying misalignment compounds. The supervisor's failure to enforce the third final warning was not just costing one wage. It was costing nine people's suppressed output. The visible cost was one. The real cost was ten.
That is the principle the Kill List operationalizes. Most organizations are paying nine-to-one cost ratios on drag they have not yet named.
Sixteen years after Bently, I was at Lockheed Martin Global Operations. I designed a program called IMPACT. The mission statement was simple. Eliminate unnecessary tasks. Enhance productivity. Bust bureaucracy.
The first time we rolled it out, five hundred employees signed up to participate inside the first twenty minutes. The performance management guide that came out of the early sessions got legs across every Lockheed division. It triggered a call from corporate. The reverse of how rollouts usually work. Bottom-up demand pulled top-down attention.
Why did people sign up that fast? Because everyone in a large organization is sitting next to work they cannot defend, work they have been carrying for years, work nobody is paid to kill. They are waiting for someone to ask the question. IMPACT was the first time most of them had heard the question.
We did not have the word Zombie Work then. We had "unnecessary tasks" and "bureaucracy." Same target. Different vocabulary. Work that does not ladder to any top objective, that survives because no one is paid to kill it, that consumes hours and cycles and senior attention and produces nothing the customer or the shareholder will ever see.
Zombie Work is the dead work that still walks. The recurring meeting that started in 2018 to solve a problem that ended in 2020 and that nobody has questioned since. The weekly status report that gets generated by twelve people and read by zero. The approval chain with seven signatures, six of which are rubber stamps that nobody is willing to remove. The dashboard that updates every Monday and is opened by no one.
10-20%
of work drives the majority of outcome. The rest is overhead, status, or drag.
The 80/20 rule has been in management literature for a century. Pareto. Juran. Drucker. Every variation says the same thing. A small percentage of activity drives a large percentage of result. The Vital Few is the AI-era operating version of that observation.
The Vital Few is the ten-to-twenty percent of work that, if done well, makes the rest of the work matter. It is the work that ladders cleanly to a top-line outcome. The Coordination Tax is what gets paid on everything that does not.
Naming the Vital Few sounds easy and is not. Most leadership teams cannot name it in a room because they have never been asked to. They can list priorities. They can list initiatives. They can list goals. The Vital Few is a smaller, sharper artifact. It is the answer to a different question. If this organization had thirty percent fewer hours next quarter and the result had to be the same, what work would survive?
That is the question the Kill List forces. Not what to add. What survives.
The Kill List audit asks three questions of every recurring work artifact in the operating model. Meetings. Reports. Approvals. Dashboards. Reviews. Steering bodies. Working groups. If the answer to all three questions is honest, the work either survives or gets a tombstone.
One. What top-level objective does this work ladder to? Not which department owns it. Which named outcome on the corporate scorecard does it move. If no senior leader can name the objective in one sentence, the work is a candidate. Most teams discover, uncomfortably, that thirty to forty percent of their recurring work cannot pass this question.
Two. What is the burden-rate cost of carrying this work for one year? Hours times loaded labor rate times frequency. Most CFOs have never seen the number for the weekly status pack. The first time they see it, the meeting either changes shape or ends.
Three. If this work disappeared on Friday, who would notice on Monday? If the answer is "no one," the work is dead. If the answer is "a leader who depends on the artifact to manage by proxy because they have lost touch with the actual signal," the work is dead and the leader needs a different solution. If the answer is "the customer, the front line, the regulator, or the board," the work is real. Only then does it survive.
Three questions. Run them against the whole calendar. The Kill List writes itself.
The structural pattern is older than AI. The nine employees at Bently in 1994 were not running an AI program. They were running a team. The math worked the same in 1994 as it does in 2026.
But AI changes the stakes. In 1994, an unnamed drag in the operating model cost you margin. In 2026, an unnamed drag in the operating model costs you the entire AI program. You cannot deploy agents against Zombie Work. The agent will dutifully automate it, and the org will keep paying for it, and the Coordination Tax will compound instead of collapse.
I have been writing some version of this thesis since 2009. The substrate changes. The principle does not. Removing structural drag so people can produce output. The vocabulary evolves. The shape holds.
The Kill List is the productized version of "fire the one, do not backfill, watch the output rise." It runs in thirty days. It produces two artifacts.
First, a quantified Coordination Tax on the P&L. Hours times burden rate times frequency on the recurring work nobody can defend. The number is uncomfortable the first time a CFO sees it. The number is actionable the second time.
Second, a named Vital Few. The work that survives audit. The work that, if done well, makes the rest of the work matter. The Agentic Pilot in Phase 2 only deploys against the Vital Few. Everything else gets a tombstone.
The structure protects the team from the addition-default. Most organizations, asked to remove work in the moment, freeze. The Kill List does not ask anyone to subtract on instinct. It runs a procedure whose only output is subtraction. The bias is bypassed by the design.
If I sat with your team next week and asked each person to write down the three recurring meetings they would kill if they had the authority, how long would the list be?
I have asked that question enough times to know the answer. The list is long. The work to write it takes ten minutes. The work to act on it takes thirty days. The math on the other side is the same math the nine employees showed me in 1994. The drag is real. The output is being suppressed. The structure can hold it, or the structure can release it. The Kill List is the release.
The Vital Few is what survives. That is the operating answer Phase 1 leaves you with. Everything from there is built on top of it.
The Readiness Score is a five-minute diagnostic across the five dimensions of agentic readiness. It tells you whether Phase 1 of the Velocity Framework is where you start, or whether you are already past it.
Take MY Readiness ScoreThree articles. One thesis. Read in order.
The Readiness Score tells you where you stand against the five dimensions that determine whether the Vital Few work is reaching execution at all.