deciqAI Strategy Execution Levers
Overview
Most organizations apply a single intervention — a town hall, a kickoff — then diagnose failure as a communication problem. The real issue: execution failure must be diagnosed across four sequential phases, each requiring a different lever. A gap at any phase propagates forward.
Cross-skill composition: Use AFTER [okr-goal-setting] (OKRs define direction; levers are the transmission mechanism). Use WITH [feedback-loops] (Lever 4 requires a designed feedback loop). Use BEFORE [resource-time-compression] (behavioral alignment must precede time compression).
When to Use
- Strategy communicated but team behavior unchanged
- Launch energy faded within 60-90 days — classic Lever 3 failure
- Post-mortem findings recur across cycles without resolution — classic Lever 4 failure
- Departments execute against different interpretations — Lever 1 not activated
- Individuals know right actions but default to old workflows under pressure — Lever 2 failure
When NOT to use: Strategy itself is wrong; org under 5 people; failure caused by resource constraints; only one phase is relevant (diagnose first, deploy only that lever).
Coaching Novices (Adaptive Front Door)
- Engine mode: user has a concrete case → run The Process directly.
- Coach mode: user is unfamiliar or says "we just need better communication" → guide step by step.
In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.
- What it is: "Most execution failures happen because organizations fix the wrong phase. This framework identifies which of four phases is broken — and what to do specifically at that phase."
- Check fit: "Has your strategy been communicated? Do people understand it? Then what happened?" If yes and execution is still failing, this applies.
- Elicit their real case: "Walk me through the last strategy initiative that didn't take hold. When did it break down — immediately after launch, in the first quarter, or when momentum faded?"
[WAIT — do not advance until user responds]
- Run The Process one step at a time — identify the broken phase first; do not move to intervention design until phase diagnosis is complete.
[WAIT — do not advance until user responds]
- Close by naming the insight: "When you fix the right phase with the right lever, the other phases that were working get amplified."
[WAIT — do not advance until user responds]
The Process
Step 1 — Phase Diagnosis
| Phase | Lever | Failure signal | Timeframe |
|---|---|---|---|
| Launch | Cognitive Alignment | Ask 5 people what the strategy means — get 5 different answers | Day 0–Week 4 |
| Kickoff | Behavior Guidance | People know the strategy but default to old workflows | Week 2–Month 3 |
| Sustained | Motivation Activation | Initial energy faded; compliance without engagement | Month 1–Month 6 |
| Retrospective | Evolution Acceleration | Same bottlenecks recur; lessons not implemented | Every cycle |
Gate: Select ONE primary phase. If multiple fail, address in sequence starting with the earliest.
Step 2 — Lever Selection
- Lever 1 — Cognitive Alignment: Co-construct strategy meaning; "teach it back" sessions. Signal: 5 people give 80%+ consistent answers. Reference: +40% comprehension vs. presentation-only.
- Lever 2 — Behavior Guidance: Redesign workflows so new behavior is path of least resistance. Signal: error rate drops across 3 consecutive periods. Reference: -65% error rate vs. training-only.
- Lever 3 — Motivation Activation: Connect individual contribution to visible impact; transparent dashboards; peer recognition tied to strategic behaviors. Signal: voluntary participation increases. Reference: +32% above-minimum engagement.
- Lever 4 — Evolution Acceleration: After-action reviews with mandatory "what will be different next cycle" commitments. Signal: time from problem to structural fix decreases cycle-over-cycle. Reference: 3x iteration speed vs. annual-review-only.
Gate: Write the specific intervention, person responsible, and measurement. If any cannot be specified, you have not selected a lever.
Steps 3-5: Design intervention (executable by someone new) → define behavioral measurement (baseline, target, timeframe) → run 4-week trial. If signal moved, continue. If not, diagnose whether intervention or phase diagnosis was wrong. Stop-rule: State "Measure [metric] from [baseline] to [target] over [timeframe] via [intervention]" — or return to Step 2.
Output: Execution Lever Diagnostic
Strategy: [one-sentence] | Launch timeframe: [X] | Failure mode: [specific]
Primary phase: [Lever 1/2/3/4] | Evidence: [observations] | Secondary: [if any]
Intervention: [executable by someone new] | Owner: [role] | Start: [date]
Metric: [behavioral] | Baseline→Target: [X→Y] | Method: [how collected]
4-week result: [value] | Assessment: [working/not/wrong phase] | Next: [action]
→ Method in Action: NASA Apollo Program Execution (1961–1969)
Execution Lever Packs
- Pack A (Tech/Product): Lever 2 most commonly fails at product-engineering handoff; keep product learning and execution learning as separate Lever 4 loops.
- Pack B (Enterprise/M&A): Lever 1 failure is most costly in transformation; Lever 3 systematically underinvested in M&A.
- Pack C (Startup <50): All four levers operate in compressed form; typical failure is over-investing in Lever 1, under-investing in Lever 4 tracking.
Applying It Well
- Diagnose the phase before selecting the intervention — communication will not fix a behavioral default.
- Fix one lever at a time, in sequence — Lever 1 must be stable before Lever 2 is deployed.
- Measure behavior, not attitude — engagement surveys are not a proxy for Levers 2 or 4.
- Lever 2 is the most underestimated lever — people default to old behavior under pressure even when they genuinely understand the new one.
- Lever 4 requires a single owner with authority to implement structural changes — not a committee.
→ Primary sources: references/sources.md
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake move | Reality |
|---|---|
| [D] "We had a great kickoff presentation, so Lever 1 is done" | A presentation to a silent audience is the input to Lever 1 — alignment requires co-construction. |
| [D] "People know what to do — the problem is motivation" | Verify Lever 2 is deployed first. Most "motivation" failures are Lever 2 gaps misread as attitude problems. |
| [D] "We did a retrospective, that's our Lever 4" | A retrospective without structural changes is a documentation exercise. Lever 4's signal is the next cycle running differently. |
| [D] "We'll roll out all four levers together in Q3" | Levers are sequential — Lever 2 before Lever 1 produces confusion. |
| [D] "Our culture already handles this naturally" | Culture substitutes for lever design only under 15 people — beyond that it preserves existing behaviors. |
| [D] "We'll track NPS / engagement to see if levers are working" | Use behavioral proxies: error rates, procedure compliance, voluntary participation — not attitude metrics. |
| [D] "Our top performers execute regardless of lever support" | Top performers mask lever failures — fragile model that breaks when they leave or stretch thin. |
| → Add [O] entries here after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags
- Same bottleneck in two consecutive cycles without structural change — Lever 4 failure
- Team members give answers to "what does this strategy mean for your work" that would lead to different actions — Lever 1 incomplete
- Strong early performance then drop after month 2-3 — Lever 3 decay signature
- "Lessons learned" documents exist but nobody can cite a resulting structural change — Lever 4 performed but not deployed
- Intervention being considered is "another all-hands" or "better slide deck" — single most common misdiagnosis
Verification
- Specific failing phase identified from observable signals?
- Single lever selected first — not all four simultaneously?
- Intervention specific enough for a new team member to execute without explanation?
- Behavioral measurement defined with baseline, target, and timeframe?
- Single named owner — not a committee? Earlier levers confirmed stable?
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/strategy-execution-levers · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/strategy-execution-levers.json