Problem Solving Manual (5): Disaggregate

The Purpose of Disaggregation

Disaggregation is the process of breaking a complex problem down into its component parts to find the specific levers of solution. By mapping the entire problem space, you can identify the critical path—the specific branches of the problem that hold the highest leverage and yield the most impactful resolutions.

The primary tool for this step is the logic tree. To ensure your logic tree is structurally sound, aim for your branches to be MECE (Mutually Exclusive, Collectively Exhaustive). This means your sub-categories should not overlap, and when taken together, they must cover the entire problem space with no gaps left behind.

Types of Logic Trees

Great problem solvers do not rely on just one type of tree; they build and iterate on multiple structures to discover which perspective yields the deepest operational insights.

1. Deductive Logic Trees

A deductive logic tree moves top-down from a known general principle or mathematical law to specific conclusions. This structure is highly certain because the underlying math or logic is absolute. A classic example is a Return on Invested Capital (ROIC) tree or the baseline financial breakdown used in our restaurant case study.

2. Inductive Logic Trees

An inductive logic tree works bottom-up, starting with scattered observations or data points and grouping them into patterns to discover a higher-level general principle. This approach is essential when you have plenty of raw data but lack a clear, overarching diagnostic framework.

Example: Inductive Logic Tree for Gia Pizza

Bottom-Up Observations (The Facts):

Daily customer count has dropped by 30% since two new restaurants opened nearby.

Many regular customers say the menu looks old and boring compared to competitors.

High-margin dishes are being ordered much less frequently.

The menu has not been updated in over two years.

Almost no marketing or promotion is being done.

New competitors are running attractive introductory offers and have modern menus.

Pattern Grouping: These facts form a clear pattern showing a drop in traffic caused by aggressive new competition, an outdated product line, and a total lack of market visibility.

Higher-Level Insight (The Conclusion): The primary root cause of the 25% profit drop is a loss of customer volume driven by stronger competition and an unappealing menu. Therefore, revenue recovery through a menu refresh and active promotions will yield the highest leverage.

3. Combining Deductive and Inductive Logic

You can work from both ends of the tree at the same time—building the trunk top-down using deductive rules while simultaneously grouping individual facts bottom-up using induction. They meet in the middle to confirm your strategic path.

Example: Combined Logic Tree

Top-Down Deductive Structure: To restore monthly profit, Gia Pizza must mathematically either Increase Revenue or Reduce Costs. Revenue breaks down further into Customers × Average Bill Size.

Bottom-Up Inductive Facts: The physical data points (30% customer drop, unchanged menu, lack of marketing) align perfectly beneath the deductive Increase Customers branch.

The Meeting Point: The highest-leverage solution is to drive customer traffic by updating the menu and running short-term promotions, while defensively negotiating lower ingredient costs to support the margin. The facts match the financial architecture perfectly.

4. Hypothesis Trees

As you gather data, you move from basic factor trees to hypothesis trees. A hypothesis tree branches out into explicit, testable claims or statements rather than simple topics. Each branch starts with a candidate solution that you actively try to confirm or disconfirm using data.

5. Decision Trees

A decision tree maps a sequence of choices where the leaves of the tree lead to concrete operational actions based on conditional rules. Not all analyses require a decision tree, but they are highly effective for mapping real-time operational procedures.

Example: Commute Decision Tree

Level 1: Is it raining?

If YES: Move to Umbrella Check.

Do you have an umbrella?

If YES: Is it rush hour?

If YES: Order a ride-share car (stay dry, tolerate slow traffic).

If NO: Walk to the train station (transit is fast, umbrella keeps you dry).

If NO: Order a ride-share car (need door-to-door transit to avoid getting soaked).

If NO: Move to Distance Check.

Is the distance under 2 kilometers?

If YES: Walk or ride a bicycle (enjoy the clear weather and fresh air).

If NO: Check traffic conditions.

If Heavy Traffic: Ride a scooter (weave through congestion easily).

If Clear Roads: Drive your own car or take a public bus.

Logical Relationships: Necessary vs. Sufficient Conditions

When structuring a hypothesis tree, the relationships between a leading hypothesis and its sub-hypotheses are governed by formal logic. Understanding these conditions helps you quickly confirm or disconfirm your strategic paths.

Necessary Conditions: These are the mandatory "must-haves." If a necessary sub-hypothesis is proven false, the leading hypothesis is completely disproven. However, proving a necessary condition true does not guarantee overall success on its own.
Sufficient Conditions: This is the threshold that acts as a guarantee. If a sufficient sub-hypothesis is proven true, the leading hypothesis is automatically proven true without needing any other evidence.

Example: Driving Profit via a New Customer Segment

The Goal: Gia Pizza successfully increases net profit by targeting families with young children.

The Necessary Conditions (The Pillars):

Market Existence: There must be a large enough population of families with kids living near the restaurant. (If there are no kids in the area, the strategy fails instantly).

Business Execution: The restaurant must have the capital to build a kids' menu or play space. (If you don't build it, the nearby families won't care).

Logical Check: Neither factor guarantees profit on its own, but failing at either one makes overall success impossible.

The Sufficient Condition (The Guarantee):

The Financial Threshold: The total incremental revenue brought in by the new kids' segment must be mathematically greater than the combined capital upgrade costs and new operational expenses.

Logical Check: Once this specific milestone is proven true, net profit increase is guaranteed. You do not need any other data points.

Pitfalls in Hypothesis-Driven Thinking

While hypothesis-driven problem solving is exceptionally fast, it is prone to critical cognitive biases if executed without guardrails:

Flawed Problem Definition: Embedding a pre-determined solution directly into the problem statement itself.
Bad: "How should we revamp our menu to bring back customers?" (This presumes the menu is the only solution).
Good: "How do we reverse the 25% profit drop?"
Narrow Framing (WYSIATI): Assuming your initial hypothesis covers the entire universe of possibilities.
Bad: "Profit is down because customers say the menu is old, so we only need to create new dishes." (This blinds you to supplier cost inflation or competitor pricing moves).
Wrong Analytical Framework: Choosing a vague tool that cannot solve the problem mathematically.
Bad: Using a generic SWOT analysis matrix to address an urgent profit leak instead of a rigorous Profit = Revenue − Cost structural breakdown.
Miscommunication Bias: Proposing a solution hypothesis simply because it is the easiest to pitch or sell to a client, rather than the one backed by core data.
Bad: "Let's just run a huge 50% discount promotion. It's simple, fast, and everyone loves a discount." (This might destroy the restaurant's remaining cash flow).
Solution Confirmation Bias: cherry-picking data to support the hypothesis you subjectively or emotionally prefer. For instance, if Dian personally loves culinary arts, she may obsess over updating recipes while ignoring supplier invoice inflation because she firmly believes "great food always wins."

The Toolkit: Cleaving Frames & Analytical Frameworks

To avoid staring at a blank page, expert problem solvers use pre-existing "cleaving frames" and analytical frameworks to chop complex situations into clean, MECE categories. However, remember that different frameworks reflect different worldviews and underlying assumptions. Avoid the dangerous trap of saying: "I have seen a problem like this before, it must be an X-type problem." Always stay ready to iterate your tree structures.

Domain	Cleaving Frames (Quick Binary Cuts)	Established Analytical Frameworks
Business	Price vs. Volume Principal vs. Agent Assets vs. Options Collaborate vs. Compete	The 3Cs (Company, Customer, Competitor) The 4Ps Marketing Mix Business Model Canvas Porter's Five Forces & PESTEL VRIO Resource Model Customer Lifetime Value (CLV)
Society	Regulate vs. Incentivize Equality vs. Liberty Mitigate vs. Adapt Supply vs. Demand	Macroeconomic Supply-Demand curves System Dynamics Modeling Socio-Economic Impact Assessment
Personal Life	Work vs. Play Short-Term vs. Long-Term Financial vs. Non-Financial	Time Management Matrix (Urgent vs. Important) Personal Cash Flow & Net Worth Statement Ikigai Framework
Operations & Finance	Fixed Costs vs. Variable Costs Capex vs. Opex Inbound vs. Outbound	P&L Cost Breakdown & Break-even Analysis ROI, NPV, and WACC calculations Operations Management Triangle (Inventory, Capacity, Variability) Economic Order Quantity (EOQ) & Newsvendor Model

For our core restaurant crisis example, we will blend three highly targeted frameworks to disaggregate the problem completely: the Profit = Revenue − Cost model, the 4Ps of Marketing (focusing on Product/Menu and Promotion), and the Restaurant Operational Model canvas.

nugistic