Consulting Prep
How to Build a Case Interview Framework from Scratch
Achraf Darkaoui · April 2026 · 7 min read
Building a conceptual framework from scratch is the hardest skill in case interviews — and the most important. If Algebra gives you equations and Process Breakdown gives you steps, Concept asks you to identify the components of a system that nobody has pre-defined for you. There is no formula to follow and no chronological order to rely on. You have to figure out the structure yourself.
This is where most candidates plateau. They can use Profit = Revenue − Cost. They can map a supply chain. But when the interviewer hands them a problem with no obvious equation or process — "Should this company enter the Indian market?" or "Why is fuel consumption increasing?" — they reach for a memorized framework that does not quite fit.
The candidates who get MBB offers do something different. They build a bespoke framework on the spot. Here is the method.
Related: Practice building conceptual frameworks with scored feedback in the CaseSights structuring drills.
What Conceptual Frameworks Actually Are
A conceptual framework decomposes a problem into distinct categories — buckets for analysis. The 3Cs (Company, Competitors, Customers) is a conceptual framework. The 4Ps (Product, Price, Promotion, Placement) is a conceptual framework. People, Processes, Systems is a conceptual framework.
What they all have in common: each one takes a system and breaks it into its distinct components.
Concepts give you buckets for analysis, nothing more. No assumptions, no hypotheses. Just the structure within which analysis happens. When you turn those buckets into hypotheses — "for this market entry to be good, what must be true about each component?" — you have moved from the MECE toolkit into structuring territory. That is a different skill.
The 3-Step Method
Step 1 — Identify the System
The system is the thing the question is about. A system is something composed of multiple parts that interact with each other. Read the question, identify the thing it asks you to analyze, and name it.
The critical rule: The system is never the metric or the outcome. It is the thing that produces the metric. Revenue is not a system — the commercial engine is. Weight is not a system — the body is. Wait time is not a system — the hospital is.
| Question | The System | |---|---| | "Should we enter this market?" | The market | | "Why are diseases rising?" | The healthcare system | | "Why is wait time increasing?" | The hospital | | "How can we win a football game?" | The football game | | "Why is fuel consumption up?" | The aircraft operation | | "Why is the production line underperforming?" | The operating model |
Once you have named the system, you are ready for step two.
Step 2 — Decompose the System into Components
There are two paths:
Path A — Import a known decomposition. If academia, consulting, or industry has already classified this system, use their work. The market decomposes into the 3Cs. The commercial engine decomposes into the 4Ps. The operating model decomposes into People, Processes, and Systems. Watch types decompose into Mechanical, Quartz, and Digital.
You are not memorizing frameworks for the sake of it. You are building a library of known decompositions that you can import when the system matches.
Path B — Build your own through mental simulation. This is the technique that changes everything. When no known decomposition fits, you play the most likely scenario in your head.
Mentally simulate how this thing most commonly works, from start to finish. Based on your own experience, your common sense, what you have observed in life. As you walk through that scenario, notice the elements you encounter — the people, the things, the conditions, the steps. Those elements become your buckets.
Example — "Why are diseases rising in this country?"
You think: people go about their daily lives, they encounter a virus, their immune system cannot fight it off, they go to the doctor, they take medication, it does not solve it, the virus spreads to others.
You extract: the virus, the population's immunity, the healthcare infrastructure, the medications available. Four distinct components of the healthcare system, each one a bucket for analysis.
Example — "Why is fuel consumption increasing?"
You think: a plane takes off, the pilot flies it at a certain altitude, the plane carries a certain weight, it follows a certain route, the engines burn fuel based on all of these.
You extract: the machine (the plane itself), the operator (the pilot), the operating conditions (weight, altitude, route). Three components that together cover the main drivers of fuel consumption.
Example — what are the components of an outfit?
You think: I get dressed in the morning. Something for my upper body, something for my lower body, shoes, then accessories.
You extract: top, bottom, shoes, accessories. Nobody taught you a "clothing framework." You just simulated the scenario and the components appeared.
The same mental process works for any problem. The scenario changes, but the technique is identical.
Step 3 — Check 80/20
Your buckets are mutually exclusive by definition — they are distinct categories. For exhaustiveness, the scenario you played in your head is your check. If your buckets cover the main elements from that scenario, you are good.
You do not need every possible bucket. You need the components that would explain the effect in the most common situation. This is the 80/20 principle applied to exhaustiveness.
Worked Example — "Should we enter the Indian market?"
Step 1: The system is the Indian market.
Step 2: We can import the 3Cs — Company, Competitors, Customers. But if we think about the scenario more carefully — what happens when a soda company enters a new market — there are more players involved. The Indian consumers, the existing soda companies, our own company's capabilities, the distribution channels, the regulators, the suppliers.
Step 3: Check 80/20. Given our understanding of the consumer packaged goods industry, the most critical components are the Indian consumers, the existing competitors, our company, and the distribution channels. Regulators and suppliers are less likely to be the deciding factor for this particular decision.
The result: a four-branch structure that is specific to this problem, built from the 3Cs but extended through mental simulation.
Why This is the Hardest Tool
For Algebra, the equation tells you the components. For Process, the chronological order tells you the steps. For Concepts, you have to identify the system and figure out its components yourself. That takes judgment, domain knowledge, and practice.
This is exactly why it is the skill that separates candidates who get offers from those who do not. Anyone can learn Profit = Revenue − Cost. Building a bespoke framework for a problem you have never seen requires a deeper capability.
How CaseSights Trains Conceptual Structuring
The structuring module of the problem-solving course is built entirely around the Concept method. It teaches the three-step process through dozens of worked examples across multiple industries and problem types. The Companion evaluates your conceptual frameworks in real-time, checking whether you identified the right system, decomposed it into appropriate components, and covered the main drivers.
Frequently Asked Questions
How do I get better at mental simulation? Practice with everyday problems. Decompose what makes a restaurant successful (the food, the service, the location, the ambiance, the pricing). Decompose what causes traffic jams (the drivers, the road design, the traffic management, the weather). The more systems you decompose, the faster your pattern recognition becomes.
What if I cannot think of the right components under time pressure? Start with the stakeholders. Who are the people and entities involved in this system? That alone gets you a reasonable first cut. Then ask yourself: are there conditions, assets, or environmental factors I am missing? This two-step fallback works for most problems.
How do I know if my framework is good enough? If an experienced consultant would look at your structure and say "yes, those are the right things to investigate," then it is good enough. You do not need perfection. You need the main components, clearly labeled, with no overlap. The 80/20 standard applies.
Should I memorize the 3Cs, 4Ps, and other frameworks? Know them well enough to import when they fit, but do not rely on them as your only tool. The value of learning the 3Cs is that when you see a market analysis question, you have a starting point. The value of the Concept method is that when you see a question where no standard framework fits, you can still build a structure from scratch.
Is the Concept tool always 80/20, never 100% MECE? Your buckets are always mutually exclusive — distinct categories do not overlap. The 80/20 applies to exhaustiveness only. In practice, perfect exhaustiveness is impossible because you could always think of one more category. Cover the main components, acknowledge what you are setting aside, and move forward.
Practice what you just read.
CaseSights drills score your structures on MECE-ness, sharpness, and specificity — in real time.
Try the Structuring Drills →