Reference · Execution
Execution — turning the strategy into delivery
Load when value-realization.
Part of the Strategy Toolkit skill · loaded on demand from SKILL.md
Six moves for carrying a strategy across the line: cut the initiative list to the few that matter, design the operating model, build the metric system, manage risk, phase and fund the transformation, and make sure the promised value is actually captured. Firm-specific frameworks are named in references/firm-lenses.md.
Initiative Prioritizer (initiative-prioritizer)
Reduce a sprawling list to the few winnable fights, sequenced, with an explicit kill list.
Method: (1) Build one register in a common value language — deduplicate; state every initiative's impact in one consistent unit (e.g. annualized profit contribution) so they're comparable, not each in its own flattering metric. (2) Plot impact × feasibility, capital as the bubble — high-impact/high-feasibility = early wins; high-impact/low-feasibility = bets to de-risk; low-impact = kill candidates. (3) Model real capacity — the binding constraint is rarely money; it's scarce leadership attention, specialist talent, change bandwidth, shared systems. State how many significant initiatives can run in parallel, justified by the binding constraint. (4) Score execution risk with DICE — Duration (interval to the next hard milestone — short beats one long run), Integrity (team/leader capability and credibility), Commitment (visible sponsorship + affected-people buy-in), Effort (added load on the people doing the work); band each likely / worrying / unlikely. A high-impact initiative with a poor DICE band is a redesign problem to fix before it starts, not a priority. (5) Cut to microbattles and sequence — a small set of specific, winnable fights each with an owner and a near-term result; map hard dependencies; sequence self-funding moves first so quick wins release the cash/capacity/proof that funds the harder battles; smooth load against the capacity number. (6) Write the explicit kill list — every initiative not done now with a one-line reason, distinguishing kill (stop) from defer (revisit at a named trigger), flagged for leadership endorsement, since the freed capacity is only real once the kills are agreed.
Output: a short roadmap of 3–5 active microbattles (each with a DICE band and a fix for any worrying band), a stated parallel-capacity number, a self-funding sequence, and an endorsed kill-and-defer list.
Done when: the active list fits the stated parallel-capacity number, and every excluded initiative carries a one-line kill or defer reason.
Long form: execution-planning in the consulting-engine pack.
Operating Model Design (operating-model-design)
Translate strategy into the few distinctive capabilities it must win on, then design around them — not boxes and lines.
Method: (1) Derive the few distinctive capabilities — restate the strategy as "we must be able to…" statements; force-rank to 3–6 that must be genuinely distinctive (a reason to win) vs the many that only need to be good enough; for each, current maturity → target → gap, and a build/buy/partner tag. If everything is distinctive, nothing is. (2) Design structure and governance from what people actually do — map the real work and cooperation before any org chart; keep the frontline close to decisions (strip layers that distance decisions from the customer); name owner and integrator roles where cooperation must happen instead of adding a coordination layer; remove before you add (every new role/layer/rule/KPI/committee must replace something). (3) Decide where two-speed fits — a fast front end (customer-facing, digital, experimentation; light governance, short cycles) connected by clear interfaces to a stable back end (core ops, risk, infrastructure); define the contract between speeds. (4) Assign decision rights on the 10–15 decisions that most determine delivery (RAPID roles: see stakeholder-alignment); stress-test for gridlock (too many Agrees) and drift (unclear Decide); push the D toward the frontline. (5) Map core processes and the breakable hand-offs — the few end-to-end flows (quote-to-cash, idea-to-launch), rank the hand-offs by how often they break and the damage, fix the worst with a seam owner and shared cross-boundary metrics before adding any gate. (6) Pressure-test coherence and sequence — capabilities, structure, decision rights, processes must point the same way; what changes day one (decision rights, owner/integrator roles) vs over the first 2–3 quarters.
Output: a one-page capability map (≤6 distinctive, each traced to a strategic choice; the "good enough" pile named), a structure that removes ≥ what it adds with named owner/integrator roles and an empowered frontline, a RAPID decision-rights map (single D per row, shortest Agree list), the fragile hand-offs fixed by owners, and a sequenced path.
Done when: no more than six capabilities are marked distinctive, every RAPID row has exactly one D, and the structure removes at least as many roles, layers or rules as it adds.
Long form: execution-planning in the consulting-engine pack.
KPI Architect (kpi-architect)
A metric system that links daily work to the strategic outcome and resists gaming.
Method: (1) Define one north-star — the single metric the strategy exists to move (test: if this improves and nothing else, did the strategy win?), with a horizon and a link to the gap to full potential. (2) Decompose into a MECE value-driver tree — the levers that sum/multiply to the north-star, no overlaps or gaps at each level, loyalty economics high in the tree (NPS → retention, repeat, referral, share of wallet are explicit drivers, not a side metric), pushed down to leaves a team can move this quarter and up to show how it closes the full-potential gap. (3) Set the three-layer metric set — outcome (north-star, slow, board-level), driver (the value-tree levers steered quarterly), health (operational leading indicators moved weekly that warn before outcomes turn); keep each layer small. (4) Set action thresholds — green / warn / action band per metric, with the named response, not just the color (a threshold with no decision attached is a number, not a control). (5) Stress-test for proxy failure and gaming — for each metric, how would a clever team hit the number without delivering the value? Pair gameable metrics with a counter-metric (volume × quality, speed × rework, survey score × actual repeat/referral); note where a metric is only a proxy.
Output: one north-star (horizon + full-potential link), a MECE value-driver tree (loyalty up front), three tight metric layers (each traceable up and down), action thresholds with named responses, a proxy-failure note + counter-metric per gameable metric.
Done when: the north-star passes the "if this improves and nothing else, did the strategy win" test, and every gameable metric has a counter-metric.
Long form: kpi-architecture in the consulting-engine pack.
Risk Mitigation (risk-mitigation)
A board-ready, strategy-scoped risk view with leading indicators that fire before each risk lands.
Method: (1) Scope risks to the strategy — only risks that threaten the strategy's outcomes and value case (not the generic enterprise catalogue), sorted into market/demand, competitive, execution, financial, regulatory, and second-order (risks your own moves create); tag each anticipate / absorb / adapt. (2) Score likelihood × impact and rank on a 5×5 grid, separating the critical few from the trivial many; note velocity (a fast risk with a slow response is more dangerous than its score). (3) Assign one owner and a specific response to the top risks (decision rights for any go/no-go: see stakeholder-alignment) — a mitigation (reduce likelihood/impact now) and a contingency (if it lands anyway). (4) Define leading indicators — the signal that moves before the risk lands (not the lagging metric that confirms it after), with a threshold, a named watcher, and a review cadence; reading weak signals early is the advantage. (5) Set governance and review cadence — hang the review on the stage gates (see value-realization) so risk status can hold up a phase release; run the pre-mortem: assume the strategy has failed 18 months out, list the ten likeliest causes without looking at the register, then check each against it and add what is missing.
Output: 12–20 strategy-specific risks (tagged by category and response class), a ranked list + heat map isolating the critical few, an owner + mitigation + contingency per top risk, a leading indicator with a threshold per top risk, a cadence wired into stage gates, and a pre-mortem that surfaced ≥1 non-obvious risk.
Done when: each top risk has an owner, a leading indicator with a threshold and a contingency, and the pre-mortem added at least one risk not already on the register.
Transformation Roadmap (transformation-roadmap)
A phased journey with owners, governance, funding, and a concrete first 90 days.
Method: (1) Phase the journey toward full potential — 3–4 phases each with a single objective (not a calendar label) and explicit, evidence-based exit criteria (phases end on evidence at a stage gate, not on dates: see value-realization); sequence so early phases de-risk and pay for later ones. (2) Place initiatives respecting dependencies and capacity, run two-speed — a fast track of visible momentum on short cycles and a stable track of foundational change (systems, structure, capability) on deliberate cycles, connected by clear interfaces; group into a few led workstreams. (3) Design the people side with Head-Heart-Hands in parallel — Head (understand the case and destination), Heart (emotional commitment, leaders role-modeling, surfacing "what it means for me"), Hands (skills, tools, processes, incentives to actually behave differently) — designed with the operational work, not bolted on as comms. (4) Design the Results Delivery Office — one place that tracks each workstream's benefit to the bottom line, runs the stage gates, holds owners to milestones, and re-scores each workstream's DICE band (see initiative-prioritizer) at each gate; score the bands now and fix any worrying or unlikely one before launch. (5) Fund the journey — self-funding sequence (see initiative-prioritizer); tie later, costlier phases' release to earlier phases hitting exit criteria. (6) Detail the first 90 days — one accountable owner (a name, not a committee) and one dated milestone per workstream, the governance cadence, early metrics, and named risks with mitigations.
Output: a phase map (objective + full-potential slice + exit criteria each), initiatives placed by phase and speed within capacity, Head-Heart-Hands moves per phase, a Results Delivery Office + DICE band per workstream, a self-funding sequence, and a one-page first-90-days.
Done when: every phase has evidence-based exit criteria rather than a date, and every first-90-days workstream has one named owner and one dated milestone.
Long form: execution-planning in the consulting-engine pack.
Value Realization (value-realization)
Make sure the value the strategy promised is actually captured.
Method: (1) Build the value ledger — every promised benefit (revenue uplift, cost reduction, capital efficiency, risk reduction) with size (annualized), owner (one accountable name), baseline, and the assumption that makes it real; reconcile the ledger total to the approved business case and to the full-potential prize — a gap means hidden optimism or double counting; resolve it. (2) Define measurement and timing per benefit — method, data source, reporting owner, and a dated ramp curve; separate planned from realized, gross from net, one-off from recurring. (3) Install results-delivery stage gates — the next phase's funding releases only when the prior phase's value is proven against the ledger (not activity-complete); gate criteria = value thresholds + evidence; a failed gate pauses/rescopes/stops (named Decider, real stop option). (4) Tie realized value to the full-potential story — map benefits to the value levers (revenue growth, margin, asset efficiency), maintain a running bridge from approved case to realized value. (5) Design the governance cadence with a reallocation rule — a value council that reviews realized vs planned; where value stalls it fixes, rescopes or stops and reallocates to compounders; it records leakage causes so the next case is less optimistic.
Output: a value ledger reconciling to the case and the full-potential prize (every line owned and baselined), measurement + dated ramp per benefit, stage gates that release on proven value with a real stop, a value bridge to full potential, and a cadence with a real reallocation rule.
Done when: the ledger total reconciles to the approved case with no unexplained gap, and each stage gate names a value threshold and a Decider with a real stop option.
Long form: kpi-architecture in the consulting-engine pack.