A KPI is defined by an outcome. A role is defined by a title. Those are two different trees, and most companies cascade the first one down the second and then wonder why a number that everybody watches has nobody behind it. The Coverage Model closes that gap with three artifacts: a Contribution Charter that states what a team owes the company, a Coverage Map that shows which capabilities are held by whom, and a Contribution Ledger that measures a person by what they moved, hold, decide and multiply rather than by the title on their contract. The title does not disappear. It stops being the thing the work is designed around.
The number everybody watches and nobody owns
Ask a management team to name the three numbers that matter most this year and you will get a fast, confident answer. Ask who owns the second one and the room slows down. Someone says the VP of Product, someone else says it is really shared with Growth, and a third person points out that the biggest lever on it currently sits inside Support. All three are correct, which is the problem.
This is not a discipline failure. It is a structural one, and it has a specific shape. A KPI is defined by an outcome: activation rate, gross margin, cycle time, renewal. A role is defined by a title: Product Manager, Ops Lead, Backend Engineer. Those two things are drawn on different axes. When a company cascades an outcome down a chart built from titles, the outcome arrives at a box that was never scoped to produce it, and the box owner does the only rational thing available: they own the part of the number that falls inside their function and treat the rest as somebody else's weather.
Melvin Conway's 1967 observation is usually quoted about software, and it applies here without modification: an organisation produces designs that copy its own communication structure. A company that draws itself in functional boxes will produce functionally boxed ownership of its own metrics. That is not a bug in the cascade. It is the cascade working exactly as designed on the wrong input.
The evidence is already pointing this way, and so is the counter-evidence
Two bodies of research matter here, and they pull in opposite directions. Both are worth taking seriously, because the design that survives contact with both is the one worth building.
Work has already left the job description. Deloitte's skills-based organisation research, based on a survey of 1,021 workers and 225 business and HR executives, found that 81% of executives reported work being performed across functional boundaries, and that a majority of the work being done falls outside people's core job descriptions. The structure of work moved. The structure of the paperwork did not.
Teams are made by how they operate, not by who is in them. Google's Project Aristotle studied 180 teams over two years across 250 attributes and concluded that how a team works together predicts effectiveness better than who is on it, including individual skill and seniority. That result is the empirical case for designing the team before selecting the people.
Now the other direction, which is the part most writing on this subject leaves out.
Ambiguity has a measured cost. Tubre and Collins' meta-analysis in the Journal of Management found a negative relationship between role ambiguity and job performance of r = -.21. That is not catastrophic and it is not nothing, and it is a real effect measured across many studies and job types. Clarity is already scarce. Gallup's 2025 engagement data found that only 46% of employees strongly agree they know what is expected of them at work, down from 56% in March 2020. And Project Aristotle, which is the headline source for the whole "how beats who" argument, ranked structure and clarity third of the five dynamics that distinguish effective teams. The same study that says the team design matters more than the roster also says the team needs clear roles, plans and goals.
There is one more finding worth putting on the table before proposing anything, because it is unflattering to the entire category. The Burning Glass Institute and Harvard Business School examined what actually happened after companies announced skills-based hiring. Fewer than 1 in 700 hires in the year studied resulted from the shift. Around 45% of firms that announced a change appear to have made it in name only, and 18% made short-term gains and then reverted. So the honest read on the language of skills-based hiring is that it has been used far more than it has been implemented. That is simultaneously a credibility problem for anyone using the vocabulary and an opening for anyone willing to run the mechanism.
The Coverage Model
The model has three artifacts and one loop. The order matters, because each artifact is derived from the one before it, and the loop is what stops the whole thing decaying back into an org chart within a year.
Artifact 01. The Contribution Charter
One page per team, written before any role is opened. It answers four questions and nothing else.
| The question | What a good answer looks like |
|---|---|
| What does this team owe the company? | Two to four outcome metrics, each with a current value and a target. Not activities. If the answer contains the word "support" or "manage", it is not finished. |
| Where does the boundary sit? | What this team decides alone, what it decides with one other team, and what it escalates. Written before the first disagreement, not after. |
| Which capabilities have to exist for those numbers to move? | A list of capabilities, phrased as verbs. This becomes the row labels of the Coverage Map, so vagueness here is expensive later. |
| Who answers for it externally? | One name. Coverage can be distributed. Accountability to the outside cannot, and pretending otherwise is how a shared number becomes an unowned one. |
Artifact 02. The Coverage Map
This is the artifact that replaces the org chart as the working document. Capabilities from the charter run down the rows. The people on the team run across the columns. Each cell holds one of three values: Primary (this person owns it and is measured on it), Secondary (this person can carry it if the primary is unavailable) or None.
Here is the example that makes the case better than any abstraction. A product team does not need a Product Manager. It needs a set of capabilities to be covered, and the number of humans required to cover them is a separate question with a separate answer.
| Capability the charter requires | Dana | Yotam | Rami | Noa |
|---|---|---|---|---|
| Talk to customers and carry the signal back | Primary | Second | None | None |
| Decide sequence against the charter metrics | Second | Primary | None | None |
| Specify a change so it can be built without re-asking | Primary | Second | Second | None |
| Build it | None | None | Primary | Primary |
| Instrument it and read what the data says | None | Second | None | Primary |
| Say no to work that does not move the metrics | None | Primary | None | None |
| Keep pricing and packaging current | None | None | None | None |
Nobody in that team carries a title called Product Manager, and the product capability is fully covered except for one row. Read the map three ways and it tells you almost everything a management review needs.
The gap
A capability with no Primary. This is the hiring spec, and it is a single line rather than twelve. Pricing and packaging is uncovered here, which is a specific, cheap thing to buy and a very expensive thing to keep ignoring.
The single point
A capability with a Primary and no Secondary. Sequencing and saying no both sit only with Yotam. If Yotam takes leave, the team keeps building and stops choosing. That is a risk register entry, not an HR note.
The stack
One person Primary on too many rows. It reads as reliability right up until it reads as burnout. Team Topologies treats this as cognitive load, and it is the constraint that quietly caps how much a team can hold.
Artifact 03. The Contribution Ledger
This is the part of the question with the least prior art, because performance systems were built on the assumption that a person maps to a role and the role maps to a rubric. When someone covers three rows of a map that no standard rubric describes, the review conversation reverts to impressions, and impressions are where bias lives.
Four entries. What you moved, what you hold, what you decide, what you multiply. Each carries a required evidence type, because an entry without evidence is an opinion with a heading.
| Entry | The question | Required evidence | How it gets gamed |
|---|---|---|---|
| Move | Which charter metric moved, by how much, and what did this person do that caused it? | The metric before and after, plus the specific intervention. Attribution stated honestly, including "contributed to" where that is the truth. | Claiming a metric that moved for market reasons. Mitigated by requiring the intervention, not just the delta. |
| Hold | How many rows are they Primary on, and what is the team's exposure if they leave tomorrow? | The Coverage Map itself. This entry is already written before the review starts. | Hoarding a capability to raise exposure. Mitigated by scoring Secondary depth as a positive, not a dilution. |
| Decide | What do they decide without asking, how reversible is it, and how far does the damage travel when one is wrong? | Named decisions from the period, placed on the reversibility and blast-radius axes. | Manufacturing irreversible decisions to look senior. Mitigated because the same axes govern who is allowed to decide alone. |
| Multiply | Who else got faster or better because of them, and how would you know? | Named people or teams, and the artifact: the runbook, the tool, the unblocked decision, the person who levelled up. | Producing documentation nobody reads. Mitigated by requiring a named beneficiary who confirms it. |
The third entry uses the same two axes as the Velocity-Risk Matrix in The Deliberate Slow: how reversible a decision is, and how wide the blast radius when it goes wrong. That is deliberate. If those axes already govern which decisions can be made fast and alone, then a person's position on them is a description of how much of the company they are carrying, and it does not require a title to express.
The legal layer, which is not optional in Israel
Any system that pays two people differently for work that looks similar has to be defensible in writing before it is used, not explained afterwards. Israel's Equal Pay for Female and Male Employees Law requires equal pay for equal, essentially equal, or comparable work. The 2020 amendment added a reporting duty that came into force for private employers with more than 518 employees from June 2022, and the required analysis is broken down by job role and ranking.
That last detail is the reason the anchor role in layer one is a structural requirement rather than a stylistic preference. The statutory unit of analysis is the role. An organisation that has genuinely dissolved roles has no denominator for its own pay report. So the model keeps the role as the reporting and market unit, and moves only the working specification into coverage. Write the framework down before applying it, apply it identically across people, document the reasoning per decision, and have an employment lawyer review it before the first pay cycle runs through it. Applied informally, a contribution system produces gaps nobody can explain later, which is both an exposure and the fastest way to lose the trust it was built to create.
How to run this without a reorganisation
None of this requires an announcement, a new tool, or a change to anyone's contract. It can be run on one team as an experiment and abandoned quietly if it does not hold.
Pick the team with the most contested number
The one where three people each own part of a metric. That is where the model has the most to prove, and where the diagnosis is fastest.
Write the charter with the team in the room
Two to four metrics, the boundary, the capability list, one external name. Ninety minutes. The argument about which capabilities belong on the list is the valuable part of the exercise, so do not shorten it.
Fill the map individually, then compare
Each person marks their own row values before seeing anyone else's. The disagreements are the finding. People routinely believe they are Primary on something the rest of the team assigned elsewhere.
Act on the three readings
Gaps become one-line hiring specs or contracted work. Single points get a Secondary named and given real practice. Stacks get rebalanced before the person tells you they are done.
Run one review cycle on the ledger in parallel
Alongside the existing process, not instead of it. Compare the two outputs. If the ledger surfaces nothing the old process missed, it is not earning its cost on this team.
Rewrite the charter at the quarter
A quarter of real measurement almost always shows that a metric was wrong or a capability was missing. A charter that survives four quarters unchanged is usually a charter nobody is using.
Where this does not fit
Three situations where the Coverage Model is the wrong tool, stated plainly because a framework that claims to fit everywhere fits nowhere.
Regulated roles with statutory duties. Where a named individual carries a legal obligation - a DPO, a money laundering reporting officer, a safety officer, a licensed professional signing off work - the role is the compliance unit and coverage cannot be distributed across it. Map the rest of the team and leave those seats alone.
Teams under about five people. Below that size everyone is Primary on nearly everything and the map is a formality. The charter is still worth writing. The matrix is not.
Companies in an acute delivery crisis. A team that is failing to ship this month needs a sequencing decision and someone to make it, not a capability inventory. Run this once the fire is out. Introducing a measurement framework during a crisis reliably produces a beautifully documented failure.
The connection to how I work
I am usually brought in against one row of somebody's map. A capability is uncovered, or its only Primary has left, and the work does not pause while a requisition is opened. So the unit I sell is the coverage rather than the seat: I hold the row for a defined period, and the engagement ends with a written handover so the row has a named Primary inside the company afterwards.
The operating baseline I run for companies is a charter and a map built across several teams at once, which is generally the first time anyone has seen the uncovered rows and the single points laid out on one page. The companion arguments are in The GEAR Model, on designing an organisation around its four needs, and The job description is a bundle, on why the requisition became the wrong unit of purchase in the first place.
Sources
- Tubre, T. C. and Collins, J. M., "Jackson and Schuler (1985) Revisited: A Meta-Analysis of the Relationships Between Role Ambiguity, Role Conflict, and Job Performance", Journal of Management, 26(1), 155-169, 2000. Source of the r = -.21 relationship between role ambiguity and job performance.
- The Burning Glass Institute and Harvard Business School, "Skills-Based Hiring: The Long Road from Pronouncements to Practice", February 2024. Source of the finding that fewer than 1 in 700 hires resulted from the shift, and of the leader, in-name-only and backslider breakdown.
- Deloitte, "The skills-based organization", based on a survey of 1,021 workers and 225 business and HR executives. Source of the 81% cross-boundary working figure. Consultancy research rather than peer-reviewed work, and read here as directional.
- Gallup, "U.S. Employee Engagement Sinks to 10-Year Low", 2025. Source of the 46% role-clarity figure and the decline from 56% in March 2020. US workforce data.
- Google re:Work, "Understand team effectiveness" (Project Aristotle). Two-year study of 180 teams across 250 attributes. Source of both the "how beats who" finding and the ranking of structure and clarity among the five dynamics.
- Strathern, M., "'Improving Ratings': Audit in the British University System", European Review, 5(3), 305-321, 1997. Source of the widely quoted formulation of Goodhart's Law.
- Skelton, M. and Pais, M., Team Topologies, IT Revolution, 2019. Source of the stream-aligned team pattern and of team cognitive load as a design constraint.
- Conway, M., "How Do Committees Invent?", Datamation, 1968. The original statement of what became Conway's Law.
- Israel Equal Pay for Female and Male Employees Law, 5756-1996, and the 2020 amendment introducing the pay-gap reporting duty for private employers with more than 518 employees, in force from June 2022. Summarised here from published Israeli employment-law commentary rather than from primary legislation, and not a substitute for legal advice.