MarTech & Customer Analytics Economics

The first comprehensive mathematical framework for CLV calculation. Systematizes approaches for computing customer lifetime value and demonstrates applications to acquisition, retention, and cross-selling decisions.

Demonstrates that CLV can value entire firms, showing a 1% improvement in retention improves firm value by 5% while the same margin improvement yields only 1%. Establishes retention's economic primacy in marketing strategy.

Operationalizes customer equity for competitive strategy, introducing the framework of Value Equity, Brand Equity, and Retention Equity. Shows how to link marketing actions to shareholder value through customer-level economics.

Comprehensive taxonomy consolidating CLV approaches for acquisition, retention, and cross-selling. Reviews probabilistic models, econometric approaches, and implementation challenges across industries.

First formal framework for the acquisition vs. retention allocation problem. Introduces the customer equity concept and demonstrates how to optimize marketing spending across the customer lifecycle.

Challenges the assumption that long-tenure customers are always more profitable. Using catalog retailer data, shows that customer lifetime and profitability are not strongly correlated—critical insight for customer selection.

Demonstrates that marketing contacts influence CLV nonlinearly across channels. Develops a framework for selecting customers and allocating resources based on future CLV rather than past transactions.

THE foundational BTYD paper introducing the Pareto/NBD model. Computes the probability a customer remains 'alive' based on recency and frequency, enabling all subsequent probabilistic customer-base analysis.

Develops the BG/NBD (Beta-Geometric/Negative Binomial Distribution) model—vastly easier to implement than Pareto/NBD while achieving similar predictive accuracy. The workhorse model for non-contractual CLV estimation.

Bridges traditional RFM (Recency, Frequency, Monetary) analysis with probabilistic CLV approaches. Shows how to use iso-value curves for customer segmentation that aligns with expected lifetime value.

The definitive overview distinguishing contractual from non-contractual settings and discrete from continuous transaction timing. Essential framework for selecting the appropriate BTYD model for any business context.

Extends BTYD models to subscription/contractual settings. Shows that ignoring customer heterogeneity leads to systematic overestimation of retention rates and customer value—critical for SaaS and subscription businesses.

Step-by-step mathematical derivation of the Pareto/NBD model with detailed exposition of the likelihood function and conditional expectations. Essential reference for implementing BTYD models from scratch.

THE paper that coined 'collaborative filtering.' Describes the Tapestry system at Xerox PARC where users could annotate documents and filter based on others' reactions—the conceptual foundation of all modern recommendation systems.

Introduces automated collaborative filtering using Pearson correlation coefficients to predict ratings. The GroupLens project at University of Minnesota became the foundation for recommender systems research.

Demonstrates that item-item collaborative filtering dramatically outperforms user-based methods in both quality and scalability. This approach became the foundation of Amazon's recommendation engine.

THE definitive paper on matrix factorization for recommendations, emerging from the Netflix Prize. Shows how to decompose the user-item matrix, handle implicit feedback, and incorporate temporal dynamics.

Introduces a probabilistic framework for matrix factorization that handles sparse data effectively through Bayesian priors. Achieves state-of-the-art performance on Netflix Prize and enables uncertainty quantification.

Comprehensive taxonomy of hybrid recommendation approaches: weighted, switching, mixed, feature combination, cascade, feature augmentation, and meta-level. Essential framework for combining collaborative, content-based, and knowledge-based methods.

Netflix's official disclosure of their recommendation architecture. Estimates that personalization saves over $1 billion annually through reduced churn—the definitive business case for recommendation system investment.

Paradigm-shifting paper showing that targeting highest-churn-risk customers for retention is often ineffective. Advocates targeting 'persuadable' customers—those whose behavior changes with intervention—rather than highest risk.

Early integration of customer lifetime duration prediction with purchase behavior modeling. Uses hazard models to jointly estimate when customers will leave and how much they'll spend while active.

Systematic comparison of machine learning methods for churn prediction. Shows that ensemble methods (bagging, boosting) substantially outperform logistic regression—establishing ML approaches for churn classification.

Develops Markov chain framework for CLV estimation when customer departure isn't directly observed. Essential methodology for subscription businesses with involuntary churn and non-contractual settings.

Demonstrates that customer churn has social network effects—when one customer churns, connected customers are more likely to follow. Quantifies the 'defection contagion' effect for retention strategy.

Comprehensive review from leading researchers on customer retention management. Covers prediction, intervention design, and optimization—synthesizing decades of research into actionable framework.

THE foundational paper for marketing mix modeling. Establishes diminishing returns, carryover effects, and competitive interactions—the theoretical underpinning of all modern MMM systems.

Pioneers distinguishing short-term versus long-term marketing effects using persistence modeling. Shows when marketing investments have permanent vs. temporary effects on sales—essential for MMM interpretation.

Proposes marginal ROMI as the unifying metric for marketing effectiveness measurement. Reviews common pitfalls in ROMI calculation and provides framework for managerial decision-making.

Technical foundation for algorithmic attribution. Develops probabilistic models for allocating conversion credit across marketing touchpoints—moving beyond heuristic rules to data-driven attribution.

Demonstrates significant biases in non-experimental advertising measurement methods through large Facebook experiments. Shows observational methods overestimate effects by 2-3x—making the case for experimental measurement.

Google's Bayesian approach to media mix modeling. Handles adstock (carryover) effects and diminishing returns through hierarchical priors—the methodology underlying modern MMM tools.

First rigorous experimental evidence on browse abandonment email effectiveness. Quantifies the causal effect of triggered emails on conversion and identifies optimal timing and content strategies.

Establishes the foundational trade-off model where personalization value must exceed privacy costs. Uses conjoint analysis to quantify consumer willingness to trade privacy for personalization benefits.

Connects personalization-privacy trade-offs to the attention economy framework. Shows how platform incentives and attention scarcity affect consumer privacy decisions and personalization acceptance.

Early experimental evidence on personalized content effectiveness. Shows how to optimize email content and timing based on individual-level response models—foundational for personalization systems.

Comprehensive survey of dynamic pricing algorithms that learn optimal prices from demand data. Covers multi-armed bandits, Bayesian learning, and exploration-exploitation trade-offs in pricing.

Decomposes promotional effects into category expansion, brand switching, and purchase acceleration. Shows how loyalty program members respond differently to promotions than non-members.

Establishes the empirical generalizations about promotional effectiveness that still guide practice today. Documents patterns in deal elasticities, stockpiling, and cross-category effects.

Develops the structural model for loyalty program ROI. Uses dynamic programming to separate true behavioral change from selection effects—essential for evaluating loyalty program investments.

Essential critical perspective on loyalty program effectiveness. Argues that most programs create loyalty to the program, not the brand—challenging assumptions about loyalty program ROI.

THE seminal paper for revenue management. Establishes theoretical foundations for dynamically pricing perishable inventory with uncertain demand—underlying all modern pricing and promotion systems.