There's a moment most managers recognize: the goal is clear (or so you think), the team is bought in (or so they say), and yet timelines slip, execution is messy, and you keep having the same conversation. The instinct is to reach for a people explanation. This team doesn't work well together. That person isn't strong enough. We need better execution.
Molly Graham's Waterline Model is a discipline against that instinct. Imagine the team as a boat: the goal is the destination, and the question is what's happening below the surface that's making the water choppy. There are four places to look, in order: structure, dynamics, interpersonal, individual.
The rule is the good part: snorkel before you scuba. Start with the shared systems — goals, roles, decision rights — before you diagnose personalities.
Structure first, because it's the most common culprit
Graham's example is a struggling marketing team she was asked to take over. Instead of assessing the people, she asked each person three questions: how would you describe your role, what goals is the team responsible for, and which number do you personally own?
The answers were wildly inconsistent. People didn't know the team's metric. Roles were defined too narrowly or disconnectedly. What leadership thought the team was for and what the team thought it was for were different things.
You cannot evaluate fit inside a broken structure, because you're measuring how people perform in a system that's pushing them in different directions. Fix the mandate, the roles, and the definition of success first — and then see who is still struggling. In her case, most of the team improved immediately, simply because they finally knew what the job was.
That's the most useful sentence in the piece: structural clarity alone resolves more issues than you'd expect. It's also the cheapest intervention, which is why it should come first.
Dynamics: behaviour that looks irrational is usually adaptive
The second layer is how the team actually works, as experienced rather than as written down. Graham's example: a founder frustrated that the team moves slowly, who also reverses decisions at the last minute and second-guesses judgment. On paper, decision rights are clear. In practice, decisions aren't stable.
The team's response is not a performance problem; it's self-protection. They slow down, add alignment layers, escalate things that don't need escalating. They optimize for not being wrong. From outside it looks like a team that can't execute. From inside it's a team that has learned what the environment punishes.
This is the same structural insight as the rest of organizational life: people adapt to the system they're in, and the system is often a set of signals rather than a set of rules. The remedy isn't a process document. It's consistency — about how decisions are made, how disagreement is handled, and when the leader intervenes. Behavior changes quickly when the signals change, because the team was already paying attention.
Interpersonal is often structure wearing a human face
Two people stepping on each other's toes is usually a role-boundary problem, an ownership overlap, or an incentive misalignment. The conflict is real, but it isn't about the relationship.
Sometimes it is. Two leaders who genuinely don't trust each other will eventually show it in slowed decisions, hoarded information, and teams quietly picking sides. Once structure and dynamics are ruled out, the work becomes direct: name the tension, tie it to the work, and set the relationship as an expectation of the role.
Individual last, and then it's a decision
If goals are clear, roles are clean, dynamics are healthy, and someone still isn't performing, you're probably looking at an individual issue. At that point the work stops being diagnostic and becomes a decision.
That ordering is the model's real contribution. It doesn't say people never matter; it says you should exhaust the shared explanations before you reach the personal one, because the shared ones affect everyone and are within your control.
Where the model can go wrong
Two failure modes are worth naming.
The first is using "it's structural" as an excuse. The model puts individual issues last, not never. A manager who never reaches the bottom layer is avoiding a hard conversation and calling it systems thinking. The test is whether the structure explanation is actually evidenced — inconsistent role definitions, unstable decisions, misaligned incentives — or just asserted.
The second is that leaders often don't control the structure. If the org design is set two levels up, the diagnostic still helps, but the remedy is escalation or adaptation, not a re-org you can't authorise.
The takeaway
The model is a cost-ordering heuristic. Shared explanations are cheap to check and affect everyone; personal explanations are expensive and affect one person. So check the cheap ones first, and don't conclude that someone is the problem until you know the system isn't producing exactly the behaviour you're frustrated by.
Snorkel before you scuba. Most of the time, the water is choppy because of the boat, not the rowers.
Source: How to debug a team that isn't working: the Waterline Model — Molly Graham, Lenny's Newsletter.