Waterline Model
The Waterline Model, developed by Molly Graham from her time at Facebook and Quip, is a framework for categorising decisions by consequence to determine how much caution and escalation they require.
The metaphor is a ship: some decisions, if wrong, cause damage above the waterline — recoverable with effort. Others punch a hole below the waterline — potentially fatal and irreversible. The discipline is classifying correctly before deciding, not after.
The two categories
Above the waterline — recoverable if wrong. Delegate broadly, decide fast. The cost of caution exceeds the cost of error; slowing down for these decisions trains teams to seek approval for things that should flow freely. Most operational decisions are above the waterline.
Below the waterline — ship-sinking if wrong. Slow down, broaden consultation, apply more analytical rigour before committing. These decisions are genuinely irreversible or have consequences too large to absorb. Strategic pivots, hiring for key roles, major architectural commitments, and public statements in sensitive situations are typical examples.
The primary failure mode
Most leaders err in the direction of treating too many decisions as below the waterline. [§ Failure modes]
This produces two downstream problems:
- The leader becomes a bottleneck — decisions queue up, and the team learns to wait rather than act.
- The team is not trained to exercise judgement — by protecting them from decisions, the leader denies them the practice that builds competence and confidence.
The cost of over-caution is invisible in any single decision but compounds over time: slower velocity, weaker teams, and a culture of escalation rather than ownership.
The classification challenge
“Below the waterline” is a judgement call that varies by company stage, domain, and available information. [§ Classification]
Early-stage: less is below the waterline because the company can absorb more variance — it is small, fast-moving, and has fewer dependencies. Late-stage: more is below the waterline because decisions have larger blast radii and more people are affected.
The failure to update the classification as the company grows is a specific error: decisions that were safely above the waterline at 10 people may be genuinely below the waterline at 500. Leaders who keep the same mental calibration across stages under-escalate on genuinely high-stakes calls.
Relationship to Jeff Bezos’s one-way/two-way door framing
The Waterline Model and the one-way/two-way door distinction are structurally similar: both sort decisions by reversibility to determine appropriate decision-making speed. [§ Related frameworks]
The practical difference: Bezos’s framing focuses on which decision to make — should we do X or not. Graham’s waterline framing focuses on who should be involved in making it — should I delegate this or escalate it. The two are complementary and operate at different levels of the decision process.