Growth Organization Design

Growth Organization Design

Design the smallest operating system that gives a named owner the levers, capabilities, decision rights, evidence, and durable support needed to move a bounded customer and business outcome. Structure follows the current work and constraint; a department name, copied org chart, headcount ratio, meeting calendar, or maturity stage is not organization design.

Read [organization-contract.md](references/organization-contract.md) before accepting reporting, ownership, staffing, performance, health, or effectiveness claims. Read [organization-methods.md](references/organization-methods.md) before choosing centralized, embedded, center-of-excellence, autonomous, service, or platform-plus-local models. Read [portfolio-and-cadence.md](references/portfolio-and-cadence.md) before prioritizing initiatives, designing forums, incentives, maintenance, or a reorganization. Read [output-contract.md](references/output-contract.md) before delivery. Use [playbook-sources.md](references/playbook-sources.md) to cite the pinned Growth Playbook basis.

Select One Mode

| Mode | Use | | --- | --- | | audit | Diagnose an existing mandate, ownership model, decision flow, team boundary, staffing pattern, portfolio, cadence, health claim, or reorganization proposal | | design | Define a new or revised outcome team, decision-rights system, capability boundary, staffing plan, operating cadence, or transition contract | | portfolio | Allocate capacity across initiatives, horizons, shared capabilities, maintenance, and strategic options using comparable evidence and explicit tradeoffs |

Name one primary mode, organization decision, accountable sponsor, customer and business outcome, current constraint as supplied or routed, scope, horizon, and external-action boundary. Public information may support dated visible role and organization hypotheses; it cannot establish private reporting lines, decision rights, staffing, performance, workload, incidents, attrition, health, or causal effectiveness.

Freeze The Organization Contract

Record product, market, business model, stage, strategy, product topology, customer state, business outcome, metric and guardrails; current constraint and its evidence; work, levers, capabilities, dependencies, shared surfaces, risks, and local context; existing teams, mandates, owners, interfaces, decision classes, escalation, staffing, management capacity, budget, portfolio, cadence, maintenance, and transition commitments; outcome, decision, execution, learning, platform, portfolio, quality, incident, workload, and sustainability evidence; sources, dates, evidence states, limitations, and requested external actions.

Use exactly verified, inferred, unavailable, or not applicable for evidence-bearing rows. A stakeholder statement without inspectable support may be a reported signal with no evidence state. Do not score unavailable organization dimensions or collapse them into a composite organization grade.

Diagnose Before Drawing The Org Chart

Use this sequence:

customer and business context
-> current growth constraint
-> bounded accountable outcome
-> required levers, capabilities, and risks
-> ownership and decision rights
-> team boundary and staffing
-> portfolio, cadence, and durable ownership
-> evidence, review, and reorganization trigger

Preserve alternatives to an organization problem: weak product value, unclear strategy, incompatible metric definitions, missing evidence, technical debt, capacity, channel economics, regulation, market context, or execution quality. Route selection of the primary business constraint to growth-diagnosis; do not invent it inside an org-design exercise.

Assign Outcome And Decision Ownership

One accountable owner must be able to explain the target customer state, business boundary, metric and guardrails, current evidence, tradeoffs, dependencies, next decision, and follow-through. A shared company goal can align functions, but each team still needs bounded controllable inputs and responsibilities. Company movement is not automatically attributable to one team.

For each decision class, name who proposes, decides, executes, reviews, can stop, and receives escalation. Distinguish reversible local decisions from shared-surface, irreversible, regulated, financial, employment, or high-risk decisions. Broad consultation is compatible with one final decider; consensus is not a substitute for authority.

Choose The Smallest Viable Model

Compare models only after the outcome and work are explicit:

| Model | Use when | Main control | | --- | --- | --- | | Central outcome team | A small number of high-value constraints need concentrated product access and capability | bounded mandate, handoff, and protection from becoming a request queue | | Embedded ownership | Customer and product context materially changes decisions and local teams have required capabilities | shared semantics, infrastructure, governance, and duplication controls | | Growth-as-a-Service | A specialist intervention is temporary and can be handed back | explicit intake, duration, knowledge transfer, and receiving owner | | Autonomous mission unit | One outcome needs dedicated surfaces, staffing, and rapid decisions | interface, maintenance, conflict, and dissolution rules | | Center of excellence | Many teams need methods, training, standards, and complex-case support | self-service default, risk-tiered review, adoption, and decision quality | | Platform-plus-local | Stable reusable capabilities serve distinct products or markets | service contracts, local configuration, overrides, escalation, and feedback |

Centralize stable repeated computation, standards, and controls when common contracts, shared quality, real consumers, reliability ownership, and lower coordination cost are evidenced. Keep customer, product, market, creative, commercial, and contextual risk judgment local when context changes the decision. Hybrid is not a conclusion; specify every interface.

Design Capabilities And Staffing

Map the work before titles or headcount. Cover only capabilities the outcome actually requires: product, engineering, data, experimentation, design, research, distribution, lifecycle, creative, sales, success, implementation, operations, finance, risk, privacy, legal, local-market, platform, enablement, and management.

For each missing capability, state work, decision rights, depth, expected load, partner dependencies, seniority, management capacity, maintenance duty, evidence-based hiring trigger, and temporary alternative. Do not use a generic “growth hacker,” copied team ratio, title prestige, or large-company benchmark as a staffing decision. Use realistic work samples that require diagnosis, evidence, tradeoffs, and a decision.

Operate A Decision Portfolio

Allocate capacity against the current constraint while preserving required maintenance and future options. Compare absolute expected contribution, eligible reach, evidence, cost, time, dependency, reversibility, risk, opportunity cost, strategic option value, and uncertainty. Scoring formulas are discussion aids, not precision.

Maintain explicit owners, evidence horizons, guardrails, dependencies, decision rules, and stop conditions. Where useful, keep core optimization, adjacent bets, exploration, shared systems, maintenance, and retirement visible as separate horizons. Reward durable value, decision quality, learning, collaboration, and system improvement; do not reward experiment wins, launches, tickets, leads, or local KPI movement alone.

Measure Organization Health Without A Composite Score

Keep these layers separate:

1. outcome: customer and business movement in the owned boundary;
2. decision: time, quality, reversal, escalation, and evidence use;
3. execution: diagnosis-to-exposure time, active work, dependency wait, and rework;
4. learning: conclusive work, decisions changed, reuse, and repeated mistakes;
5. platform: reliability, adoption, self-service, queue, support, and duplication;
6. portfolio: capacity by outcome and horizon, stopped work, maintenance, and retirement;
7. sustainability: incidents, workload, on-call, meeting load, attrition, and key-person risk.

Protect employee privacy and interpret people metrics cautiously. Output volume, attendance, speed, or high usage cannot prove effectiveness and may coexist with harm, lock-in, debt, or poor decisions.

Reorganize Only From A Decision-Flow Trigger

Valid triggers include an unowned dominant constraint, repeated stable work worth productizing, a central queue blocking contextual decisions, duplicated local foundations creating risk, product or market portfolio divergence, a PMF or strategy change, or rising maintenance and trust debt. Slower growth, a missed quarter, leadership preference, or maturity aspiration alone is insufficient.

Before recommending change, define which decisions improve; alternatives; affected work and people; transition cost; what stops; unfinished-work, system, and customer ownership; permissions and knowledge transfer; communication; review date; success and harm measures; and adjustment or rollback conditions. Preserve mandate stability long enough to execute and learn.

Route Specialist Work

Route primary constraint selection and enterprise execution routing to growth-diagnosis; model structure to growth-model-design; metric semantics to growth-measurement; event implementation to growth-measurement; test assignment, power, analysis, and causal readout to growth-measurement; technical shared-capability architecture and sourcing to growth-operations; customer evidence to customer-research and positioning; product value evidence to product-market-fit-assessment; and reward mechanics to growth-loop-design.

This Skill owns organization contracts, outcome and decision rights, model selection, functional boundaries, staffing scenarios, initiative portfolios, operating cadence, maintenance ownership, health evidence, and reorganization specifications. It does not replace employment, legal, labor, compensation, security, privacy, finance, or local-market review.

Deliver In Order

Return:

1. mode, decision, sponsor, outcome, constraint state, scope, horizon, and action boundary;
2. organization contract and evidence ledger;
3. current decision-flow, capability, dependency, and ownership map;
4. model alternatives and central-versus-local decision record;
5. outcome ownership and decision-rights matrix;
6. team, role, staffing, interface, maintenance, and transition specification;
7. prioritized portfolio and operating cadence;
8. disaggregated health scorecard, risks, review triggers, handoffs, and pinned Playbook sources.

For China work, keep market, language, locale, product, customer, channel, creative, lifecycle, pricing, identity, consent, payment, invoice, partner, support, privacy, risk, regulatory, and operating conditions separate. Do not infer any platform, provider, law, employment, payment, consent, or organization condition from translation or geography alone.

External-Action Boundary

This Skill creates and reads local artifacts only. Do not access HRIS, payroll, recruiting, performance, calendar, communication, project, analytics, experiment, warehouse, CRM, finance, identity, customer, or cloud systems; export employee or customer data; change reporting lines, employment, permissions, goals, headcount, compensation, hiring, budgets, project allocation, schedules, ratings, systems, traffic, or customer state; contact employees, candidates, vendors, or customers; publish; procure; deploy; or claim a reorganization is complete without separate task-level authorization and required reviews.

Keep One Output Language

Use the requested output language consistently across headings, prose, tables, labels, and actions. When no language is explicit, match the user's dominant language; market, locale, platform, and source language do not override it.

For Simplified Chinese, write natural Simplified Chinese and translate ordinary business or analytical jargon instead of embedding English words such as owner, brief, listing, cohort, baseline, benchmark, guardrail, gate, finding, roadmap, workflow, and handoff. Keep only proper names, standard acronyms after a Chinese first-use definition, machine tokens or IDs, code, formulas, filenames, URLs, and exact quotations where necessary.

For English, use idiomatic English and do not add Chinese glosses except for proper nouns or quoted source text. Use multiple languages only when explicitly requested, and keep each version in a separate labeled section rather than mixing languages within sentences or tables. Do not alternate languages for emphasis or perceived expertise.

Completion Gate

Confirm that the decision, outcome, constraint state, product and market context, required work, levers, capabilities, dependencies, risks, owner, decision classes, functional boundaries, model alternatives, staffing assumptions, management capacity, portfolio, maintenance, cadence, health layers, evidence states, transition costs, review and reorganization triggers, specialist handoffs, privacy boundary, pinned sources, and external-action limit are explicit; shared goals have controllable responsibilities; consensus does not replace authority; activity is not called value; no universal org chart, headcount ratio, maturity target, or composite score was invented; and no external action occurred.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: krillinai
• Source: krillinai/growee-skills
• License: MIT
• Homepage: www.clawee.ai/

Install and usage instructions live in the source repository linked above.