Formulas & Cheatsheets

Product Definition: A product delivers value to a user segment. Categories include physical (manufacturing bounds, depreciating inventory), digital (infinite replication, marginal cost ~0), AI-driven (probabilistic outputs, non-deterministic scaling), and hybrid (e.g., IoT connecting physical sensors to digital dashboards).

Brand Man Concept: Developed by Neil McElroy at P&G in 1931. McElroy proposed that individual products should have dedicated strategic owners ('Brand Men') who manage product positioning, sales, and strategy end-to-end, serving as the modern precursor to the Product Manager ('CEO of a product' without formal authority).

Product vs. Project: Product Management is outcome-focused (Why/What, user retention, lifetime value, iterative changes). Project Management is transaction-focused (How/When, scope, timeline, budget, hitting fixed milestones).

SWOT Framework: Maps internal attributes (Strengths, Weaknesses) against external environmental conditions (Opportunities, Threats) to plan market positioning.

Ansoff Matrix: Four-way growth model: Market Penetration (sell existing product to existing market), Market Development (sell existing product to new market), Product Development (sell new product to existing market), and Diversification (sell new product to new market - highest risk).

PESTEL Scan: Macro environmental scanning tool mapping Political factors (tariffs, regulations), Economic (inflation, interest rates), Social (consumer demographics, lifestyles), Technological (GenAI adoption, computing costs), Environmental (sustainability constraints), and Legal (data privacy, antitrust rules).

Porter's Five Forces: Audits industry profit potential via: 1) Threat of new entrants (capital barrier size), 2) Buyer power (low switching costs increase this), 3) Supplier power (supplier consolidation increases this), 4) Threat of substitutes (indirect alternatives), and 5) Competitive rivalry.

Blue Ocean Strategy & ERRC: Blue Ocean focuses on creating uncontested market spaces, rendering competition irrelevant. The ERRC grid helps break the cost-value trade-off by identifying factors to: Eliminate (industry standards to cut), Reduce (factors to scale down), Raise (factors to increase value), and Create (brand new factors to introduce).

SCAMPER Ideation: Substitute, Combine, Adapt, Modify, Put to other use, Eliminate, and Reverse. Lateral thinking checklist to brainstorm feature upgrades.

Nominal Group Technique: A structured brain-writing process where participants write ideas silently, present them in turn, discuss, and execute blind voting to select priorities, neutralizing groupthink and dominant voices.

Spotify India Case Study: Spotify entered a highly competitive, pirated, and multi-lingual Indian music market. They succeeded by localizing recommendations across 8+ regional languages and introducing micro-payments (sachet pricing: e.g., daily passes for ₹7) to align with credit-card-averse youth.

Jobs-to-be-Done (JTBD): Customers do not buy products; they 'hire' them to achieve progress in specific circumstances. Understanding the customer's functional, social, and emotional job enables better feature scoping than simple demographic targeting.

Value Proposition Canvas (VPC): A tool under the Business Model Canvas. It details the customer profile (Jobs, Pains, Gains) and maps it directly to the value map (Products & Services, Pain Relievers, Gain Creators) to verify product-market fit.

Zerodha Case Study: Zerodha built a massive market moat by offering a zero-brokerage model for long-term investing and flat-fee trading. They eliminated expensive human relationship managers (RMs), relying entirely on automated online onboarding and self-serve dashboards, which scaled profitably via volume.

Feasibility Dimensions: Desirability (Do customers want this? tested via mockups/surveys), Feasibility (Can we build this? checked via technology constraints), and Viability (Should we do this? assessed via revenue potential and cost structures).

Market Sizing: TAM (Total Addressable Market - total global market demand), SAM (Serviceable Addressable Market - portion of TAM targetable with your business model/channel), and SOM (Serviceable Obtainable Market - portion of SAM you can realistically capture in the short term).

Platform Moats: Platform business models connect multi-sided markets (e.g. hosts and guests). They scale non-linearly using network effects: direct same-side effects (more users attract same users, e.g. WhatsApp) and indirect cross-side effects (more hosts attract more guests, e.g., Airbnb). Liquidity acts as a massive competitive barrier.

Design Thinking (5 Steps): Empathize (understand users via research), Define (synthesize insights to draft core problem statements), Ideate (explore design variations), Prototype (build low/high-fidelity mocks), and Test (observe usability logs).

Swiggy Case Study: Real-time mapping solved the emotional anxiety of hungry users who wanted delivery status visibility, embodying design thinking's empathy and visibility principles.

Personas: Evidence-based user representations. Primary Persona: target user whose needs are non-negotiable and dictate the core interface flow. Secondary Persona: complementary user whose needs must not disrupt the primary user's experience.

Avika Case Study: A mental health app that mapped custom UX paths based on stress levels. High-anxiety users were shown simplified card interfaces with direct buttons to reduce cognitive overload, while low-anxiety users had access to detailed tracking tabs.

User Journeys & Stories: Journeys map user actions, feelings, and friction points across touchpoints. User Stories capture feature value: 'As a [user persona], I want [action], so that [benefit].'

MoSCoW Prioritization: Must Haves (critical path MVP), Should Haves (important but not critical), Could Haves (optional/nice to have), and Won't Haves (deferred for future release cycles to prevent scope creep).

RICE Score: Formula: (Reach * Impact * Confidence) / Effort. Reach is users/month, Impact is score value, Confidence is percentage data proof, Effort is person-months. Scores features objectively to bypass the HIPPO effect.

Prototyping Fidelity: Low-fidelity (paper sketches, rapid layout tests, cheap modifications) vs. High-fidelity (interactive Figma mocks, represents visual guidelines and click paths, used for client checks before coding).

Jakob Nielsen's Usability Heuristics: 10 guidelines for UI design, including: 1) Visibility of System Status (keep user informed), 2) Match between System and Real World (use familiar terms), 3) User Control and Freedom (undo/redo functions), 4) Consistency and Standards (uniform behaviors), and 5) Error Prevention (checking format on the fly and warning before actions).

Double Diamond Model: Innovation framework: Discover (diverge - user research) → Define (converge - project scope) → Develop (diverge - design templates) → Deliver (converge - release production).

Client-Server Model: Separates front-end UI presentation layer (client) from backend processing, logic, and data storage (server) communicating via secure API channels (e.g. REST, GraphQL).

Microservices Architecture: Deconstructs monolithic code into decoupled, independent services (e.g., search, billing, shipping). If the payment microservice fails, the search microservice stays online. This enables modular scaling and faster deployments.

CI/CD Pipelines: Continuous Integration automates compiling code and running unit tests as developers commit changes. Continuous Deployment automates deploying the verified build to staging/production, minimizing human configuration bugs.

DevOps: Synthesizes development and operations, using pipeline automation and server health metrics to speed up shipping quality releases.

Moravec's Paradox: Computers easily execute high-level logical reasoning and math (which humans find difficult), but struggle with basic sensory and motor tasks (which humans perform intuitively like picking up items or face recognition).

Learning Paradigms:

• Supervised Learning: Trained on labeled inputs and outputs (e.g. classifying spam).
• Unsupervised Learning: Finds patterns in unlabeled data (e.g. clustering user segments).
• Reinforcement Learning: Agent learns via trial and error inside environments to maximize rewards.

Deep Learning: Neural networks with hidden layers. Training adjusts weights via backpropagation, which calculates output errors and distributes adjustments backward. This parallel arithmetic requires GPUs (Graphic Processing Units) for acceleration.

Transformer Architecture: Replaced recurrent models (RNNs/LSTMs). Uses a self-attention mechanism to read all input words in parallel, calculating context weights for word-pair connections instantly. Pre-training compresses huge text databases into model weights (lossy pre-training compression).

RLHF Alignment: Reinforcement Learning from Human Feedback aligns raw text-predictors to act as helpful, safe, and conversational assistants rather than simply predicting the next token.

AI Agent Definition: An autonomous software system that perceives its environment through sensors, processes logic, and interacts via actuators to achieve goals.

PACE Design Framework: Performance metrics (safety, speed), Environment (domains), Actuators (API hooks, wheels), and Sensors (text, cameras).

Agent Types: Simple Reflex (condition-action rules), Model-Based (keeps internal world model to track unobserved factors), Goal-Based (seeks targets), Utility-Based (optimizes score desirability using utility functions), and Learning Agent (improves behaviors over time).

System 1 vs. System 2: System 1 is fast, reactive next-token prediction. System 2 is slow, analytical planning (reflection loops, chain-of-thought, tool calls).

RAG (Retrieval-Augmented Generation): Mitigates model hallucinations by fetching relevant factual text chunks from local vector databases based on query embeddings and appending those facts into the LLM context prompt.

n8n Workflows & Docker: n8n connects APIs, databases, and AI nodes visually. Script execution nodes (Javascript/Python) are run inside sandboxed Docker containers to prevent custom code from accessing host system files (unlike heavy VMs, Docker containers share the host OS kernel, keeping overhead low).

Agent Memory: Short-term (session histories), Long-term (vector database indices), Episodic (chronological interactions), and Semantic (factual facts).

AI Agent Security: Jailbreaking/Prompt Injection (tricking LLMs into bypassing safety guardrails via inputs) and Data Poisoning (inserting malicious or false files into training data or RAG vector indexes to distort outputs).

Methodologies: Waterfall (sequential: Requirements → Design → Code → Test → Deploy; failed in the CityCare case because late-stage integration testing failed to handle complex, shifting hospital APIs). Agile (iterative, value-driven, incremental releases).

Scrum Pillars: Transparency (shared standards), Inspection (regular checkups), and Adaptation (process adjustments when deviations are found).

Scrum Roles: Product Owner (owns the backlog and prioritizes features), Scrum Master (obstacle removal, coaches team, facilitates syncs), and Developers (commits to and codes the sprint increment).

Scrum Ceremonies: Sprint Planning (sets sprint backlog), Daily Scrum (15-min daily alignment stand-up), Sprint Review (presents product increment to stakeholders), and Sprint Retrospective (process evaluation and team improvement).

Backlog items: Epics (massive strategic blocks of work), User Stories (specific requirements), and Tasks/Bugs (individual work cards).

INVEST Story Criteria: Independent, Negotiable, Valuable, Estimable, Small, and Testable.

Fibonacci Estimation & Planning Poker: Relative sizing in story points (1, 2, 3, 5, 8, 13) instead of hours. Prevents arguments and reflects that larger tasks carry exponentially higher uncertainty.

Velocity: Average story points completed per sprint, used to predict future team capacities.

Jira & JQL: JQL is Jira Query Language used to filter issues. Example: project = 'Mobile' AND status = 'Ready for QA' AND assignee = currentUser().

Interest-Influence Matrix: Maps stakeholders. High Interest/High Influence (manage closely), Low Interest/High Influence (keep satisfied via high-level milestones, avoiding overloading them with daily operations).

Lead Without Authority: Product managers do not manage engineers directly. They lead using active listening, data proof, and structured communication models.

Storytelling Frameworks:

• SCQA: Situation (context), Complication (friction trigger), Question (challenge), and Answer (solution proposal).
• STAR: Situation, Task, Action, and Result. Effective for highlighting project achievements.

Business Ethics: Managing data privacy, auditing algorithmic bias in models, and mitigating carbon footprints in green data centers.

Go-To-Market (GTM) Situational Analysis (5 Cs): Customer (needs, segments), Company (internal capabilities), Competitors (market share), Collaborators (suppliers, partners), and Context (macro PESTEL trends).

STP Strategy: Segmentation (splitting the market), Targeting (focusing on subsets), and Positioning (brand differentiation message).

Marketing Mixes: 7 Ps (Product, Price, Place, Promotion, People, Process, and Physical Evidence). 7 Ms for campaign execution (Market, Mission, Message, Media, Money, Measurement, and Material - the creative ad files).

Kapferer's Brand Identity Prism: 6 facets: Physique (visual identity), Personality (character), Culture (values/roots), Relationship (connections), Reflection (how the user wants to be seen), and Self-Image (internal feeling when using the brand).

Customer Decision Models:

• AIDAS Funnel: Attention → Interest → Desire → Action → Satisfaction.
• FAB Framework: Features (attributes) → Advantages (differentiators) → Benefits (user outcomes).
• Maslow's Hierarchy: Aligning product messaging with psychological needs.

Search Marketing: SEO (organic ranking, search keyword matches) vs. SEM (paid Google search ads, CPC bid structures).

GEO (Generative Engine Optimization): Optimizing text and website data structure (such as structured tables, citations, and summaries) to ensure AI generative engines quote your brand when answering customer queries.

Atomberg Case Study: Disrupted the home fan market using energy-efficient BLDC fans and scaling via direct-to-consumer (D2C) social media marketing loops.

Growth Loops: Cyclical engines of user acquisition where actions of active users naturally feed back to acquire new users (e.g. virality, referral credits). Replaces traditional linear acquisition funnels.

Pricing Approaches: Cost-based (markup on margins), Competition-based (matching competitor pricing), and Customer-value-based (pricing based on user outcomes/ROI - standard for SaaS).

New Product Pricing: Market Skimming (high initial prices for early adopters, lower over time; e.g. new iPhones) vs. Market Penetration (low initial prices to build volume quickly and activate network effects; e.g., streaming apps).

Price Elasticity of Demand (PED): % Change in Quantity Demanded / % Change in Price. Bounded: PED > 1 is elastic (sensitive to price), PED < 1 is inelastic (price insensitive).

Distribution Channels: Direct (selling D2C online) vs. Indirect (using retailers, distributors). Coverage: Intensive (available everywhere), Selective (filtered retail partners), and Exclusive (single distributor).

RATER Service Quality: Reliability, Assurance, Tangibles, Empathy, and Responsiveness.

Cognitive Biases: Confirmation Bias (looking only for data confirming pre-existing assumptions, ignoring negative logs) and HIPPO Effect (Highest Paid Person's Opinion overriding empirical database data).

Research Types:

• Exploratory Research: Qualitative methods (user interviews, focus group discussions) to discover and define customer problems.
• Descriptive Research: Quantitative methods (customer surveys) to audit market segment sizes.
• Causal / Experimental Research: Quantitative testing (A/B testing, price adjustments) to establish concrete cause-and-effect relationships.

Goodhart's Law: 'When a measure becomes a target, it ceases to be a good measure.' If a team optimizes clicks, they might trigger confusing prompts, driving up click rates but destroying session retention.

Linear Regression: Models relationships between independent predictors (ad spend, web speed) and a dependent variable (daily sales). Standard equation: Y = b0 + b1*X1 + b2*X2 + error.

R-squared (R²): Coefficient of Determination. Measures the percentage of variance in the dependent variable explained by the independent predictors. R² ranges between 0 and 1.

Significance: Bounded by the p-value. A p-value < 0.05 indicates statistical significance (less than 5% chance the observed relationship is random noise). If F-significance p-value > 0.05, the regression model is statistically untrustworthy, even if R-squared is high.

Dummy Coding: Converting categorical variables (e.g. Device type: iOS, Android) into binary dummy codes (0 or 1) so Excel's regression tool can process them.

Product Metrics: North Star Metric (the single metric capturing core customer value delivery; e.g. hours watched on Netflix) vs. Vanity Metrics (impressive numbers that do not reflect retention; e.g. total app downloads).

User Journey Metrics: Adoption (initial signups) → Engagement (DAU/MAU ratios, session length) → Retention (cohort retention curves over time) → Monetization (ARPU, LTV).

Cohort vs. Funnel Analysis: Cohort analysis tracks retention of specific groups over time (e.g., users signing up in Jan vs. Feb). Funnel analysis maps drop-off percentages at each transaction stage (e.g., Landing → Cart → Checkout → Success).

A/B Testing: Causal experiment. Randomization is critical to distribute confounding variables evenly between cohorts, isolating the change as the sole cause of differences. p-value < 0.05 indicates statistical significance.

Netflix Shakespeare Case Study: Netflix automated experimentation using their internal 'Shakespeare' platform, allowing PMs to configure user splits, segment data, and review automated p-values without writing SQL scripts.