
Problem-Solving
This framework was generated using the Framework Builder methodology. It demonstrates the five-layer DNA architecture applied to problem-solving and root-cause diagnosis.
Built with the Framework Builder methodology
Download markdownThis framework was generated using the Framework Builder methodology. It demonstrates the five-layer DNA architecture applied to problem-solving and root-cause diagnosis.
Layer 1: Principles Foundation
Principle 1: The name of the problem is usually a symptom. "Churn," "delays" and "low quality" are labels. Each hides several distinct causes that need different fixes.
Principle 2: Size before you solve. Segment the problem by impact, age and source before investigating. A cohort that never got started is a different problem from long-standing accounts that suddenly leave.
Principle 3: The constraint is the leverage point. If one failure mode produces most of the damage, fixing it beats improving six minor ones by ten percent.
Principle 4: Diagnose before you prescribe. Spend a meaningful share of your time budget, around 30%, on understanding the problem. Action feels efficient and often fixes the wrong thing.
Principle 5: Define "solved" up front. Name the metric, the size of the change and the time frame before anyone starts fixing.
Layer 2: Systematic Approach
Step 1: Validate the problem statement. Ask what the person actually needs. "The report is slow" may mean "I need this data live."
Step 2: Triage. Is it an acute spike or a chronic pattern? For a spike, check external triggers: a price change, a release, a competitor move, a billing error. For a chronic pattern, segment by cohort and find which segments drive the trend.
Step 3: Branch on data. If usage or process data exists, correlate it with the problem to find the behaviors that predict it. If data is thin, run structured interviews with a fixed question set.
Step 4: Map each segment to a cause. Typical categories: onboarding failure, feature gap, support quality, pricing mismatch. Write the ruled-out causes next to the chosen one.
Step 5: Fix, then check recurrence. Apply the fix to one segment, then watch whether it holds for 90 days.
Layer 3: Force Multipliers
Multiplier 1: Reframe the question. Replace "how do we reduce X" with "what would need to be true for the person in month three to choose to stay." It turns prevention into value creation and surfaces different causes.
Multiplier 2: Sort constraints into real, assumed and self-imposed. Assumed ones, usually phrased as "people just expect that," are the most fertile ground.
Multiplier 3: Keep a pattern library. Tag every resolved case by type, root cause and solution. Check it before starting a new investigation.
Multiplier 4: Use the escape hatch. If testing a plausible cause takes thirty minutes, try it before building a three-day analysis.
Multiplier 5: Record what you ruled out. Alongside the chosen cause, write the causes you considered and the evidence that excluded them. The next investigator will not repeat the work.
Layer 4: Success Metrics
Positive indicators:
- The target rate falls for the segment you fixed, and stays down at 90 days
- Time from noticing a problem to naming its cause shrinks as the library grows
- New problems start from a known pattern more often
- A fix holds and a different problem surfaces elsewhere (you solved the real cause and can move to the next)
Failure signals:
- The fix works for a month and the number returns (symptom, not cause)
- Solutions are proposed in the first meeting
- No one can say what "solved" meant
- Every problem is investigated as if it were new
- Analysis time exceeds the cost of simply trying the likely fix
Layer 5: Implementation Guidance
Start with one segment. Choose the one with the greatest impact and the most available data. Run the full process on it alone.
Log every decision point. What data you pulled, how you segmented, where you branched and which causes you ruled out.
Add each solved case to the pattern library. After three to five cases, recurring patterns start to speed up diagnosis.
Common mistake: solving the stated problem. You get a perfect answer to a question nobody needed answered.
Edge case: no data at all. Run five structured interviews before choosing a direction, and record the questions so they can be reused.
Edge case: two plausible causes, one cheap test. Test the cheap one first and write down the result either way.
Common mistake: jumping to solutions. Proposing fixes in the first meeting feels like action and often addresses the wrong thing. Hold the diagnosis block of time before any fix is discussed.
Common mistake: treating every problem as unique. Roughly 70% of problems in any domain are variations of ones you have seen. Without a pattern library, you pay the discovery cost every time.
Common mistake: no definition of "solved." If you cannot say what metric should change, by how much and by when, the work never closes and you cannot tell when you failed.
Common mistake: over-investing in analysis. When two plausible causes exist and one test is cheap, run the test. Keep the escape hatch in the framework, and use it on purpose.
Built with the Framework Builder methodology. Get the skill →