Stop Calling, Start Targeting: Why Strategy and EDA are the Engines of Marketing Machine Learning

2026-02-13 · Source: Machine Learning on Medium · Field: Business & Management — Marketing, Branding & Advertising, Corporate Strategy & Leadership · Depth: Intermediate, short

Summary

Effective machine learning in marketing, exemplified by a Portuguese bank's campaign, hinges on robust exploratory data analysis (EDA) and clear business strategy, rather than isolated model building. A baseline conversion rate of 11.7% highlighted the need to move beyond "blanket marketing" models. The case study demonstrated that addressing data integrity issues, such as remapping "unknown" previous outcomes to "no previous contact" for first-time prospects, transformed a seemingly faulty variable into a primary model driver. Furthermore, identifying and removing data leakage, like call duration predicting success post-sale, was crucial for building a predictive model useful before customer contact. The analysis also emphasized balancing model evaluation between resource efficiency (precision-focused) and market penetration (recall-focused) to align with specific business objectives, moving beyond the "accuracy paradox" where high accuracy can be commercially useless.

Key takeaway

For Data Scientists developing marketing models, you must prioritize comprehensive exploratory data analysis and align model objectives directly with business strategy. Ignoring data integrity issues or the "accuracy paradox" can lead to commercially useless models, even if statistically accurate. Ensure your models support decisions *before* an action, such as a call, by carefully identifying and removing data leakage.

Key insights

Effective marketing ML requires deep EDA and clear business strategy to avoid optimizing for wrong metrics or using leaky data.

Principles

• Data integrity precedes imputation or deletion.
• Model evaluation must align with business goals.
• Remove features that leak future information.

Method

Interrogate missing data to uncover business insights, remap ambiguous categories, identify and remove data leakage, and apply balancing techniques like SMOTE for imbalanced datasets.

In practice

• Cross-reference missing data with other features.
• Evaluate models based on precision or recall.
• Use SMOTE for imbalanced classification tasks.

Topics

• Marketing Machine Learning
• Exploratory Data Analysis
• Data Integrity
• Model Evaluation
• Class Imbalance

Best for: Machine Learning Engineer, Data Scientist, Marketing Professional

Related on AIssential

• Your Model Scored 96% Accuracy. Think That’s Success? Think Again. — Model accuracy alone is insufficient; calibrated probabilities and cost-optimized thresholds drive real business value.
• What Data Modeling Is and Is Not — Effective data modeling is crucial for business operations, preventing data disasters, and enabling both human and machine decision-making.
• Why Powerful Machine Learning Is Deceptively Easy — Methodological rigor in ML is crucial to distinguish genuine model performance from illusory gains.
• Key Strategies for Building Effective Machine Learning Systems — Effective ML system development prioritizes iterative refinement, data quality, and appropriate model complexity for robust, real-world performance.
• The $200 Billion Data Debt: Every Major AI Lab Is Running Out of Fuel and Nobody Is Talking About… — The AI industry faces a critical data debt crisis as high-quality human-generated data is exhausted and synthetic data causes model collapse.
• Why Your 99% Accurate Model Might Actually Be Useless — Data leakage inflates model accuracy by providing unavailable information, leading to real-world failure.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning on Medium.