Skip to content

Toolkit — Minto pyramid

Gate: G2 Decompose. Category: structure substrate.

What problem it solves

A framed problem is one paragraph; the decomposition needs several layers of structure, and those layers have to cohere. The Minto pyramid forces the decomposition to start from a single governing idea and branch downward into mutually exclusive, collectively exhaustive supporting groups — so that every later gate reads the tree as a whole, not as a set of unrelated bullets. Without the pyramid, G2's tree collapses into a list that no subsequent gate can route against.

How it is used

A 2-hour G2 workshop, chaired by the G2 lead, with the engagement team and one outside reader. The group starts from the framed problem paragraph and writes the pyramid's top — the one sentence the engagement's answer will eventually support. Then, working downward, the group names supporting groups (usually 2–4 at each level) and recurses until each leaf is sharp enough to route at G3.

Inputs

  • Framed problem paragraph (the question the pyramid will answer).
  • System description (so the decomposition is grounded in real decision points).
  • The team's initial, messy bullet list of what might be in the engagement's scope — the raw material the pyramid will structure.

Outputs

  • A pyramid: one governing sentence at the top, 2–4 supporting groups beneath, each recursively supported by 2–4 sub-groups, down to routable leaves.
  • A MECE check-list (covered separately): a note per level confirming that the groups are mutually exclusive and collectively exhaustive, with any overlaps or gaps flagged.
  • Group-level summary sentences: one sentence per group at every level, which is what a G3 reader uses to route.

Visualisation

Minto pyramid — one top-level sentence supported by three second-level groups each supported by two to three leaves Top — governing answer one sentence; what the engagement concludes Group A supporting reason Group B supporting reason Group C supporting reason A.1 leaf A.2 leaf B.1 leaf B.2 leaf C.1 leaf C.2 leaf One top, three groups, six leaves — each layer MECE against the layer above.

The pyramid is a tree with the answer at the root and routable leaves at the bottom. Each parent's supporting children must together fully justify the parent.

Anatomy

Top. One sentence — the answer the engagement will eventually give. At G2 the top is provisional (the engagement's hypothesis about its own answer); it is updated if the tree's branches produce a different answer. What matters is that the tree has a root to decompose against.

Groups. The direct supports of the top. Minto's rule: 2–4 groups at each level. Fewer than 2 and there is no decomposition; more than 4 and the reader cannot hold the groups in mind. If the engagement produces 7 groups, it has skipped a level.

Recursion. Each group has the same structure: a summary sentence and 2–4 supporting children. The recursion stops when a leaf is sharp enough that G3 can route it against the five routing question marks.

Summary sentences. Each non-leaf node has a one-sentence summary — what the whole subtree below it says. These are what a later-gate reader uses to understand the tree without reading every leaf.

Example

Paper trail — building the freight-yard pyramid

G2 workshop, 100 minutes. Team: Priya, Raj, Jin (dock solver lead), Amira (escalation lead). Chair: Ada. Outside reader: Marco (regional ops director).

T+0 — top sentence. From the G1 frame: "Unloading times have grown from 28 to 44 minutes; drivers wait; carriers leave." Ada: "the governing answer is probably about where the time went. Draft top: The 16-minute growth is concentrated in three bottlenecks: slot allocation, dock-door coordination, and carrier-ETA misalignment."

T+25 — second level. Three groups: slot allocation, dock-door coordination, carrier-ETA. Are they MECE? Ada: "anything left out?" Raj: "driver-side causes — drivers arriving without paperwork." Ada: "that's a carrier-ETA sub-case." Kept as a sub-group.

T+55 — third level. Slot allocation has three leaves (assignment delay, reassignment churn, override handling). Dock-door coordination has two leaves (operator handoff, door-clash). Carrier-ETA has three leaves (ETA accuracy, paperwork arrival, priority conflicts). Eight leaves total.

T+75 — outside reader. Marco reads the top-down summary. Marco: "you've split carrier-ETA and slot allocation as separate groups, but priority conflicts in carrier-ETA drive reassignment churn in slot allocation. Overlap." Ada: "yes — that's a MECE issue. Either the two groups merge, or the overlap is named as a cross-cut." After 10 minutes of discussion, the team records priority conflicts as a cross-cut with a note — it is reviewed in both groups but counted only once in the total scope.

T+100 — close. Pyramid photographed. Group summaries written. Eight leaves plus one cross-cut. The leaves feed the G3 routing map.

Pitfalls

More than 4 groups. A level with 7 groups is a level that has skipped its own structure. Find the missing meta-category.

Groups without summary sentences. A group named "carrier issues" has no summary — it is a topic, not a claim. Each group must say something.

Top sentence that is a question. The top is the answer, not the question. "What is driving unloading delay?" is the engagement's framing; the pyramid's top is "Unloading delay is driven by three bottlenecks…" — the answer the tree will support.

Leaves too coarse or too fine. Too coarse: the leaf is still a subject, not a piece G3 can route. Too fine: the leaf is an implementation detail that belongs in G4, not G2. The chair's test: can we ask the five routing questions of this leaf, and get distinct answers?

Pyramid as the plan. The pyramid decomposes the problem, not the work. The decomposition hands routable leaves to G3; it does not prescribe the order or the team.

When not to use

  • Engagements with one routable piece. The pyramid is a single branch; the work is not worth the workshop.
  • Pure-research engagements where the answer structure is genuinely unknown and the decomposition is premature. The pyramid can be written after initial investigation, not before.

Provenance

Minto's Pyramid Principle [1] — derived from her work at McKinsey — is the canonical treatment. The same underlying tree-structured decomposition is recognised in the issue-tree and MECE traditions (both covered separately in this toolkit) and predates Minto in the operations-research literature on hierarchical decomposition [2].

  • Issue tree. The pyramid's tree-shaped cousin; used when the top is a question rather than an answer.
  • MECE. The check run against each level of the pyramid.
  • Framed problem paragraph. Supplies the top sentence's input.

Verification

[1] Minto B. The Pyramid Principle: Logic in Writing and Thinking. 3rd ed. Pearson; 2009. [verified] The canonical treatment of pyramid-structured decomposition.

[2] Simon HA. The architecture of complexity. Proceedings of the American Philosophical Society. 1962;106(6):467–82. [verified] The foundational treatment of hierarchical decomposition in complex systems.