A meta-framework connecting decision tools into a complete lifecycle. Different stages require different tools—using the wrong tool at the wrong stage wastes effort or creates false confidence.


Quick Reference

I’m stuck at…Use thisCore question
”What am I even deciding?”Can I state this in one sentence?
”Should I even think about this?”Can I undo it? What’s the real cost?
”What do I need to know?”What could go wrong? What does good look like?
”How do I find answers?”Where does borrowed signal live?
”Have I researched enough?”Pattern or outcome uncertainty?
”Am I ready to act?”Can I complete the sentence?
”What should I actually do?”Minimum test that produces signal?
”Why am I not acting?”Can I point to what I did today?
”What does this result mean?”Process or outcome? Me, approach, or context?
”What do I do next?”Execution failure or strategy failure?

The Sequence

1. ROUGH FRAME    → "What am I deciding?" (one sentence)
      ↓
2. WEIGH          → "Does this need the full process?"
      ↓
3. FULL FRAME     → Generate the questions that matter
      ↓
4. RESEARCH       → Gather borrowed signal efficiently
      ↓
5. CHECK LIMIT    → Is my uncertainty about pattern or outcome?
      ↓
6. PREPARE        → Can I complete the sentence for each outcome?
      ↓
7. DESIGN         → What's the minimum experiment to learn?
      ↓
8. ACT            → Convert clarity to action same-day
      ↓
9. INTERPRET      → What does the signal I received mean?
      ↓
10. DECIDE        → Iterate, pivot, or quit?

The loop continues: Iterate returns to DESIGN, Pivot returns to FRAME, Quit exits.

Why ROUGH FRAME before WEIGH: You can’t assess reversibility until you know what you’re deciding. “Is this reversible?” requires knowing what “this” is.


Stage 1: ROUGH FRAME

Question: “What am I deciding?”

State the decision in one sentence. Just enough to evaluate whether it needs the full process.

Not “should I change careers?” but “should I apply to this specific job?” Not “how do I fix my finances?” but “should I cancel this subscription?”

The rough frame scopes what WEIGH evaluates.


Stage 2: WEIGH

Question: “Does this decision need the full apparatus, or should I just try it and see?”

“Is the cost of analysis higher than the expected cost of being wrong?”

ReversibleIrreversible
Low cost if wrongJust try itLight process
High cost if wrongLight processFull process

When you take the simple path:

  • Set a cheap fail-fast condition: “If X doesn’t happen in Y time, I’ll reassess”
  • You still know your fallback: “If this doesn’t work, I’ll Z”
  • You’re not skipping preparation—you’re betting that learning from outcome beats learning from research

See Simplicity Moves Cost, It Doesn’t Reduce It — taking the simple path is valid when the cost is survivable and you’re the one bearing it.


Stage 3: FULL FRAME

Question: “What do I need to know to make this decision well?”

Use Question Generation SOP:

  1. What does good look like? (Success criteria)
  2. What could go wrong? (Failure modes)
  3. What don’t I know that I need to know? (Knowledge gaps)
  4. Who has done this before? (Sources of borrowed signal)

The output is a list of questions to answer before acting.


Stage 4: RESEARCH

Question: “How do I efficiently find answers before hitting the limit?”

Where borrowed signal lives:

Question TypePrimary SourcesSecondary Sources
”What typically happens?”Case studies, statistics, precedentExpert opinion, forums
”What do experts recommend?”Practitioner interviews, guidelinesBooks, courses, blogs
”What are the failure modes?”Post-mortems, reviews, complaintsYour pre-mortem thinking
”What’s the process?”Official documentation, regulatoryHow-to guides, tutorials
”What’s the cost/benefit?”Comparable transactions, market dataEstimates, rules of thumb

Minimum viable search:

  1. One expert source (someone who’s done this repeatedly)
  2. One case study (someone who’s been through this)
  3. One failure mode search (what goes wrong?)

If these don’t exist → you may have already hit the limit.

Diminishing returns signals:

Old SignalProblemBetter Signal
Same info, different wordsSEO makes this default”New perspectives or new presentations?”
Circular citationsNormal in most fields”Traced back to primary sources?”
Tangential driftCould be productive”Connect to core question in one sentence?”

Productive Divergence vs. Avoidance Drift:

ProductiveAvoidance
Explicitly in EXPLORE modeShould be in DECIDE mode
Can articulate why this connectsStruggle to explain why here
Set a time-boxLost track of time
Return to core questionForgot original question

Stage 5: CHECK LIMIT

Question: “Is my remaining uncertainty about the general pattern, or about my specific outcome?”

Use The Limit of Borrowed Signal:

“Can more research tell me what will happen to ME, or only what typically happens to PEOPLE LIKE ME?”

  • Pattern uncertainty → Keep researching (borrowed signal helps)
  • Outcome uncertainty → Stop researching, engage to get personal signal

The limit is reached when your uncertainty is downstream of an interaction that hasn’t occurred yet. No amount of research can tell you how this specific person will respond to your specific pitch.


Stage 6: PREPARE

Question: “Can I complete the sentence for each plausible outcome?”

Use The Preparedness Threshold:

“If this outcome happens, then I take this specific action using this resource, which results in this expected result.”

If you can complete this sentence for the range of plausible outcomes—good, bad, and ambiguous—you’re prepared. If you can’t, identify what’s missing and address it.

Preparedness isn’t certainty about what will happen. It’s clarity about what you’ll do when you don’t know.


Stage 7: DESIGN

Question: “What’s the minimum engagement that produces useful signal?”

Preparedness ThresholdExperiment FrameOutput Contract
”I know what I’ll do if it fails""I’ve designed this to teach me either way""I acted today”
DefensiveGenerativeActivating

Before acting, ask:

  1. What am I testing? (State the hypothesis)
  2. What result would confirm vs. disconfirm? (Define success signal)
  3. What’s the minimum cost/effort to learn this? (Scope the experiment)
  4. What will I do with each possible result? (Pre-commit to interpretation)

“Most things in life aren’t pass or fail—they’re experiments you can adjust.”

This reframe increases rigor by forcing you to define what you’re learning, not just what you’re doing.

Needs development

The experiment design process could use more structure—specifically, templates for common experiment types (sales outreach, job applications, creative pitches, technical tests).


Stage 8: ACT

Question: “Did I convert clarity to action today?”

Use The Output Contract:

“Can I point to what I did today that I wouldn’t have done without this thinking?”

If not, the thinking was either premature or performative.

Clarity is fuel. Action is fire. Without output, planning reinforces impossibility by inertia.


Stage 9: INTERPRET

Question: “What do I actually learn from this result?”

Step 1: Separate outcome from process

“Even if the outcome had been different, would I make the same choice again given what I knew?”

Good ProcessBad Process
Good OutcomeSkill (repeat)Luck (don’t over-learn)
Bad OutcomeVariance (continue)Failure (learn and change)

Step 2: Locate the signal

“What’s the simplest explanation for this result?”

Signal TypeWhat it tells youWhat to change
About the contextThis situation, timing, counterpartyTry different context, same approach
About your approachYour method, pitch, strategyRefine or redesign
About youYour skills, preparation, fitDevelop, practice, or reposition
About the goalWhether this is achievable/worth itReassess goal itself

Start with context, then approach, then you, then goal. Don’t jump to “I’m fundamentally unsuited” when “their budget got cut” explains it.

Step 3: Determine update size

FactorUpdate moreUpdate less
Sample sizeMultiple consistent signalsSingle data point
Signal qualityDirect feedback, clear reasonVague rejection, no explanation
Base rate deviationVery different from typicalMatches base rates
Process qualityGood process, bad outcomeBad process (outcome uninformative)

Pre-Committed Interpretation: The best time to decide how to interpret a result is before you get it—while still in DESIGN. This prevents motivated reasoning, overreaction, and goalpost-moving.

Needs development

Could use concrete examples of pre-committed interpretations for common scenarios.


Stage 10: DECIDE

Question: “Given what I learned, what do I do next?”

OptionWhen to useSignal pattern
IterateApproach directionally rightPartial success, clear feedback on what to improve
PivotGoal valid but approach isn’t workingRepeated failure despite good execution
QuitGoal isn’t achievable or worth the costFundamental obstacle, or opportunity cost too high

Test 1: Execution vs. Strategy

“Am I failing at execution (how I’m doing it) or failing at strategy (what I’m doing)?”

  • Execution failure → Iterate
  • Strategy failure → Pivot or quit

Test 2: The Viability Question

“What would have to be true for this to eventually work?”

If the conditions seem impossible or not worth creating → Pivot or quit

Test 3: The Sample Size Question

“How many attempts before I conclude this isn’t working vs. just bad luck?”

  • Highly variable (sales, creative pitches): 10-20 attempts
  • Moderately variable (applications, outreach): 5-10 attempts
  • Low variability (technical tests): 3-5 attempts

Legitimate Quit Conditions:

  1. Fundamental obstacle discovered: Makes this impossible, not just hard
  2. Better opportunity emerged: What I’m not doing is clearly better
  3. Cost exceeded budget: Hit my limit on time/money/energy
  4. Goal changed: I no longer want what I thought I wanted

The test: “Am I quitting because I’m discouraged, or because the evidence says this won’t work?”


When to Build Systems

Not every repeated action deserves a system. The investment must match the problem.

The “No System” Test:

If…Then…
Task happens < 10 times/monthManual is probably fine
Errors don’t compoundManual with checklist
Data isn’t needed for future decisionsDon’t capture it
You’re the only userYour memory IS the system

The Data Capture Anxiety Test:

“If I don’t have this data in 6 months, what decision will I be unable to make?”

If you can’t name the decision → you don’t need to capture it.

Trigger Thresholds (at least ONE must be true):

TriggerThreshold
Repetition> 10x/month
Error rate> 5% with consequences
Time drain> 5 hours/week
Scale inflectionGrowth makes manual unsustainable
Data criticalityCompliance or decision-critical
Stakeholder frictionRepeated complaints

If no triggers met: Continue manual, reassess monthly.

Build, Buy, or Outsource:

Score each question 1-4 (Rare → Common):

  1. How widespread is it in its market?
  2. How well do people understand it?
  3. What are publications saying about it?
  • 3-6 points: Build it (no viable alternative)
  • 7-9 points: Buy it (solutions exist)
  • 10-12 points: Outsource it (commodity)

North: Where this comes from

East: What opposes this?

South: Where this leads

  • Faster decisions (right tool, right stage)
  • Better learning (designed experiments, pre-committed interpretation)
  • Transferable process (hand off without draining myself)

West: What’s similar?