library / lib0fdde36cafb26f93
Predictive Marketing: Easy Ways Every Marketer Can Use Customer Analytics and Big Data
Omer Artun, Dominique Levin · 2015
In a sentence
A practical guide showing how every marketer—not just data scientists at giant companies—can use big data, customer analytics, and machine learning to deliver personalized experiences that maximize customer lifetime value.
Predictive Marketing demystifies big data and machine learning for everyday marketers, arguing that the same predictive analytics that powered Amazon, Netflix, and Harrah's are now cheap, accessible, and easy to deploy. Authors Ömer Artun and Dominique Levin lay out a complete primer on how predictive analytics works under the hood, how to build complete customer profiles, and how to manage customers as a value portfolio. They then deliver nine concrete 'plays'—from optimizing marketing spend and predicting customer personas to launching predictive programs that convert, grow, and retain customers—each grounded in real company stories and measurable returns. The book's central promise is to restore the personal, one-to-one relationships of the old corner store at the scale of millions of customers, creating a win-win for customers (relevant experiences), businesses (profitable relationships), and marketers (career-making visibility).
The four lenses
- Science
- Statistics
- Systems
- Strategy
The model
A causal model in which data and predictive analytics capabilities (design levers) enable personalized, relevant customer experiences (psychological/behavioral states) that increase customer lifetime value and enterprise value (outcomes), moderated by customer value segment and life cycle stage.
Customer Data Integration & Qualitydesign lever
The organizational and technical capability to collect, clean, deduplicate, and link all customer data (online and offline, transactional and behavioral) into a single complete 360-degree customer profile available in near real time.
Predictive Intelligence Capabilitydesign lever
The capability to apply predictive analytics models—clustering, propensity/likelihood models, and recommendation engines—to customer data to generate forward-looking insights about customer behavior, value, and preferences.
Campaign Automation & Executiondesign lever
The capability to trigger and orchestrate personalized customer experiences and campaigns across channels (email, web, social, mobile, direct mail, call center, store) driven by predictive insights in near real time.
Marketing Relevance & Personalizationpsychological state
The degree to which marketing communications, offers, and recommendations match an individual customer's needs, preferences, context, and life cycle stage, delivering personalized rather than one-size-fits-all experiences across touch points.
Customer Engagementbehavioral pattern
The level and consistency of a customer's active interaction with the brand, including email opens and clicks, website visits, purchases, social interaction, and product usage, reflecting the strength of the customer-brand relationship.
Customer Trust & Loyaltypsychological state
The customer's accumulated confidence in and emotional attachment to the brand, built through valuable, respectful, and consistent experiences, which makes customers more likely to return, spend more, and advocate for the brand.
Likelihood to Buy/Engagebehavioral pattern
The predicted probability that a given customer or prospect will make a purchase or engage with the brand in a future period, used to rank customers and prioritize marketing investment and discount levels.
Customer Value Segmentcontextual condition
The classification of a customer into high-, medium-, or low-value tiers (e.g., top 10%, next 60%, bottom 30%) based on lifetime value, which moderates how marketing and retention resources should be allocated to that customer.
Customer Life Cycle Stagecontextual condition
The stage of a customer's journey with the brand—prospect, new, repeat/active, at-risk/inactive, or lapsed—which determines the appropriate marketing objective and strategy and moderates the effect of campaigns on outcomes.
Customer Retentionoutcome metric
The degree to which customers continue to do business with the brand over time, measured preferably by dollar-value retention, and including the prevention of churn and value migration (declining spend among retained customers).
Customer Lifetime Valueoutcome metric
The total revenue or profit, adjusted for costs and time-value of money, expected from an individual customer over the duration of their relationship with the brand, treated as the single most important metric in marketing.
Enterprise Value / Customer Equityoutcome metric
The total firm or shareholder value driven by the sum of the lifetime values of all customers (customer equity), which the book argues is maximized by maximizing each individual customer's lifetime value.
How they connect
- customer data integration → predicts predictive intelligence
- customer data integration → influences marketing relevance
- predictive intelligence → predicts likelihood to buy
- predictive intelligence → influences marketing relevance
- campaign automation → influences marketing relevance
- likelihood to buy → influences campaign automation
- marketing relevance → predicts customer engagement
- customer engagement → predicts customer trust loyalty
- marketing relevance → mediates customer trust loyalty
- customer trust loyalty → predicts customer retention
- customer engagement → predicts customer retention
- customer retention → predicts customer lifetime value
- customer engagement → influences customer lifetime value
- customer lifetime value → predicts enterprise value
- customer value segment → moderates customer retention
- life cycle stage → moderates marketing relevance
A candidate measure
Predictive Marketing: Easy Ways Every Marketer Can Use Customer Analytics and Big Data — derived measurement candidates
Customer Data Integration & Quality
% customers with single 360 profile; Duplicate/dedup rate; In-store email/data capture rate; Profile update frequency
self-report suitability: medium
Predictive Intelligence Capability
Number/type of models in production; Model lift vs RFM/baseline; Self-learning refresh cadence
self-report suitability: medium
Campaign Automation & Execution
Count of triggered/automated campaigns; Number of coordinated channels; Use of holdout/A-B tests
self-report suitability: medium
Marketing Relevance & Personalization
Consumer-reported relevance; Open/click/conversion lift personalized vs generic; Recall of relevant campaigns
self-report suitability: high
Customer Engagement
Email open/click counts; Web/app session counts; Days between orders; Distinct categories purchased
self-report suitability: low
Customer Trust & Loyalty
Repeat purchase rate; Referral counts; Loyalty program participation; Trust/advocacy survey scores
self-report suitability: high
Likelihood to Buy/Engage
Propensity decile rank; Predicted probability score; Realized purchase vs prediction
self-report suitability: low
Customer Value Segment
Percentile rank by spend/profit; Annual spend tier; Profitability after returns/discounts
self-report suitability: none
Customer Life Cycle Stage
Days since first/last purchase; Order count; Subscription active/expired flag
self-report suitability: none
Customer Retention
Retention/churn rate by segment; Dollar-value retention; Net active customer change
self-report suitability: low
Customer Lifetime Value
Historical LTV (margin minus acquisition cost); Predicted future value; Share of wallet/upside
self-report suitability: none
Enterprise Value / Customer Equity
Sum of customer LTVs; Total revenue and margin; Firm valuation
self-report suitability: none
The story
The reader An everyday marketer at a company large or small who wants to deliver relevant, personalized experiences and earn respect and visibility within their organization.
External problem
Customer data is scattered, messy, and inaccessible, making it impossible to deliver relevant marketing or maximize customer lifetime value.
Internal problem
The marketer feels overwhelmed by big data and machine learning, fearful of math, and worried about falling behind early adopters.
Philosophical problem
One-size-fits-all mass marketing is disrespectful to customers and just plain wrong; customers deserve relevant, meaningful experiences.
The plan
- Build complete, accurate 360-degree customer profiles by integrating and cleaning your data.
- Analyze customers and assess value at micro and macro levels using predictive models.
- Manage customers as a value portfolio across the life cycle.
- Run the nine easy plays to acquire, grow, and retain customers.
- Choose the right technology and start small, then iterate.
Success
- Customers receive relevant, delightful experiences and stay loyal.
- The business builds more profitable customer relationships and grows enterprise value.
- The marketer gains visibility, respect, and career advancement as an early adopter.
At stake
- Customers feel ignored, opt out, and defect to competitors offering personalization.
- Marketing spend is wasted on irrelevant, one-size-fits-all campaigns.
- The company falls behind competitors who have already adopted predictive marketing.
Related in the library
- Internet Marketing Integrating Online and Offline Strategies
- Hacking Growth: How Today's Fastest-Growing Companies Drive Breakout Success
- Internet Marketing Made Easy
- Internet Marketing Metrics: The 8 Most Important Metrics to Track
- Lean Analytics: Use Data to Build a Better Startup Faster
- Marketing Management