Molly Graham on Giving Away Your Legos, the Waterline Model, and Scaling Yourself

Molly Graham has led people and culture at Facebook, Quip (acquired by Salesforce), and Lambda School. She is known for her writing on scaling organisations and has coached leaders across dozens of high-growth companies. Her essay “Giving Away Your Legos” has become a canonical reference in startup management.

Key ideas

Giving away your Legos: as companies scale, individuals must willingly hand off work they love to new hires. Those who hold on create bottlenecks; those who let go earn the trust needed to do more important work.
The J-Curve of team change: performance drops whenever a company reorganises, scales, or brings in a layer of management. The dip is predictable and temporary. Leaders who understand the pattern do not panic; those who do not understand it fire people or reverse course too early.
The Waterline Model: four layers — structural, dynamics, interpersonal, intrapersonal — must be worked through in order. Most coaching fails because it goes straight to the intrapersonal (“you need to be more self-aware”) before fixing the structural problems that make self-awareness irrelevant.
Six goal rules: goals should be few, shared, measurable, binary, ambitious, and time-bound. Nearly every company violates at least three on the first try.
The Glue Club: the people who create alignment, unblock others, and hold cultural context across an organisation are essential at every stage. They are systematically undervalued by companies that measure only individual output.

Episode content

Giving away your Legos

Graham’s original essay described the experience of watching someone else do the work you used to love — and having to actively choose to be glad about it rather than resentful. The Lego metaphor captures the emotional reality: you spent years building something, you know every brick, and now someone else is building the next section.

The choice to give away Legos is not just philosophical; it is practically necessary. A person who holds on to every brick they owned at 50 people will become a bottleneck at 500 people. The bottleneck is invisible at first — they are just “staying involved” — and becomes obvious too late.

The deeper insight is that giving away Legos is what earns the trust to do harder work. Leaders who demonstrate they can let go are the ones who are handed the most consequential responsibilities as the company grows.

The J-Curve

Every significant org change — a new management layer, a merger, a reorg, a fast hiring burst — produces a temporary performance dip. This is the J-Curve: performance falls from a previous stable point, reaches a low, then recovers higher than before if the change was correct.

The dip is not a sign that the change was wrong. It is a sign that people are adapting. The danger period is the bottom of the J: a leader who does not understand the pattern interprets the dip as failure and either reverses the change or replaces the people who are struggling to adapt.

Graham’s advice: tell people about the J-Curve before the dip arrives. When they experience the dip, they can recognise it as predicted rather than alarming. Naming the pattern gives people a frame for their experience.

The Waterline Model

Graham uses a diving metaphor: you can snorkel on the surface or scuba dive to the bottom. Most coaching drives straight to the bottom (intrapersonal: “you need to work on your self-awareness”). The Waterline Model argues that before going deep, you must check each layer in order.

Layer 1 — Structural: does the person have the authority, resources, and role clarity to succeed? If a person cannot make decisions without twelve approvals, no amount of self-awareness will fix their effectiveness.

Layer 2 — Dynamics: are the team rituals, communication patterns, and meeting structures working? A person can be personally excellent in a team that lacks the processes to make good decisions together.

Layer 3 — Interpersonal: are the relationships between specific people creating friction? This is the layer most managers jump to (“they just don’t get along”) without first checking whether the structural and dynamic layers are healthy.

Layer 4 — Intrapersonal: is there a fundamental belief, mindset, or behavioural pattern in the person that is the root cause? This is rarely the first problem; it only becomes the problem once the upper layers are clear.

The practical instruction: before giving feedback at any layer, work your way down from the top. Snorkel before you scuba.

Six goal rules

Graham’s rules for effective goals: they should be few (a person can hold three to five in mind; more dilutes focus), shared (everyone on the team owns them, not siloed), measurable (unambiguous signal at the end of the period), binary (you either hit the goal or you did not — no partial credit, which removes the incentive to sandbag), ambitious (goals that are obviously achievable before the period starts are not goals, they are a list of planned activities), and time-bound (a goal without a deadline is a wish).

The most common violations: too many goals (people feel productive listing them; focus requires cutting), binary becomes fuzzy (leaders accept 80% as success), and ambitious becomes sandbagged (“we always hit our goals” is not a compliment).

Escalation

Graham argues that escalation is a tool, not a failure. When someone cannot resolve a conflict with a peer, escalating to a shared manager is the correct move — not a sign of weakness. The failure mode is people who spend months unable to resolve a conflict because they think escalating is embarrassing. A manager who receives a proper escalation has useful information. A manager who finds out six months later that two team members were stuck has been failed by the culture, not served by it.

The Glue Club

Graham identifies a category of high-value workers she calls “glue”: people who connect teams, keep projects moving, resolve ambiguities, and carry organisational knowledge across department lines. Glue work is rarely captured in performance reviews, which measure individual output. People who do glue work well often receive average reviews despite being essential.

The Glue Club failure mode: someone is doing glue work instead of the work in their job description. This happens when there is a gap in the org — no one is doing the glue — and the person who notices is the one who fills it. Over time they are recognised for their individual output (which is low, because they are doing glue) rather than their contribution (which is high). The fix is either to restructure so someone’s job is explicitly glue, or to make the glue contribution visible.