HOW RETAILERS CAN LEVERAGE DATA TO STAY COMPETITIVE IN AN EVER-CHANGING DIGITAL LANDSCAPE Luca Piccolo | Manager Michele Miraglia | Manager
AGENDA 1 Introduction 2 Retailers landscape 3 Data-driven value cases 4 Prenatal Retail Group: a practical example 5 Key learnings & takeaways
INTRODUCTION
BIG DATA & VISUALIZATION 200+ Big Data Engineers & Data Scientists Architects and developers with wide experience in big data platforms, cloud 50+ Production projects, up and running & real-time and visualization tools Speech @STRATA in NY: «...turning Data into value » (IOT) DATA SCIENCE Scientists specialized in designing and implementing Advanced Analytics solutions, ML, AI UK FOCUS GROUPS London Strong vertical domain knowledge and DE Düsseldorf experience, dedicated consultants on IT Munich Turin IoT platforms, Agriculture Market, and Milan Quantum Computing Rome
ADVISO ISORY & & ED EDUCA UCATIO TION FACTOR ORY & & DELI ELIVER ERY Advisory to assist and drive company Project management, designing and implementation data trasformation in order to assess data , technology professional services to enable ideas and prototypes and human capital with the purpose of designing to become a data-driven product . business case, processes and organization This process is characterized by agile development step and data driven decision system The Data Incubator Machine Learning Training course for employees Models building and industrialization and graduated student to develop to deploy predictive analytics Data Science competences on new in production environment generation analytical tools The Data Lab Big Data Platform Consulting and advisory service which Integration and development of advanced allows to drive data experimentation analytics solutions to support business that unlock business value decisions and actions
RETAILERS LANDSCAPE
SOME KEY ELEMENTS OF THE LANDSCAPE Digital and The concept Speed Omnichannel Data physical of store & customized protection
DATA-DRIVEN VALUE CASES
Functional units WHERE DATA VALUE LIES SALES & MKTG DISTRIBUTION PRODUCTION OUR EXPERIENCE SUPPORTING RETAILERS SERVICE CUSTOM CUS OMER R DIM DIMENS NSIO ION PRODUCT PR ODUCT DIME DIMENSI NSION ON Customer Logistics Prioritisation Optimization Production Understanding Optimization & Targeting Service Price Improvement Tuning
LOGISTICS OPTIMIZATION DIMENSION: PRODUCT - FUNCTION: DISTRIBUTION APPLICATIONS VALUE CASES WHY? Sales forecast Sales forecast Stock-out and over-stock Predictive demand planning & stock optimization reduction & strategic planning Revenue/space increase Layout optimization Product placement optimization Replenishment planning Cost reduction Distribution network optimization Distribution optimization Predictive demand & production planning
PRODUCTION OPTIMIZATION DIMENSION: PRODUCT - FUNCTION: PRODUCTION APPLICATIONS VALUE CASES WHY? Quality prediction Product quality increase Automatic quality drop & waste detection Waste cost reduction Waste root cause analysis Waste cost reduction Early anomaly detection Suppliers evaluation Maintenance planning Predictive maintenance Downtime reduction Maintenance cost reduction
PRICE TUNING DIMENSION: PRODUCT - FUNCTION: SALES & MARKETING APPLICATIONS VALUE CASES WHY? Customized pricing Margin optimization Dynamic pricing Marketing automation Customized promotions Campaign automation Discount & margin optimization Phase-out tuning Over-stock reduction Improved price understanding Price prediction & tuning Product features value inference Support in pricing new products
CUSTOMER PRIORITISATION DIMENSION: CUSTOMER – FUNCTION: SALES & MARKETING APPLICATIONS VALUE CASES WHY? Customer lifetime value Value drop detection Customized promotions Upselling Recommendation support Churn prediction Increased retention Customized promotions Campaign optimization Engagement campaigns
UNDERSTANDING & TARGETING DIMENSION: CUSTOMER - FUNCTION: SALES & MARKETING APPLICATIONS VALUE CASES WHY? Single customer view Enable up & cross-selling Customized marketing actions Improve customer service level Omnichannel interaction Online journey optimization Proactive customer support Funnel optimization Most searched / viewed Real-time pop-ups Physical journey tracking Layout optimization Data-driven product placement Cross-selling Customized real-time campaigns Recommendation & Next Best Cross-selling & upselling Marketing automation Offer Customer engagement Coupons & banners
SERVICE IMPROVEMENT DIMENSION: CUSTOMER - FUNCTION: SERVICE APPLICATIONS VALUE CASES WHY? Text-based feedback analysis Trend detection Targeted actions Topic analysis Service chat analysis Churn prediction Real-time customized actions
A PRACTICAL EXAMPLE
BUSINESS SCENARIO 4 Specialized Brands 2 for pregnancy and childcare 2 for toys More then 700 POS in Europe (300 in Italy)
BUSINESS SCENARIO Until 2015 the four brands were controlled from different companies and were competitors In 2017 M&A operation brings all the brands within the control of a single private company: Artsana S.p.A. Each brand has it’s own positioning , commercial strategy, customer base and tone of voice
BUSINESS NEEDS Know your customer Know your product • • Most of the customers Most of the products buy in different brands are common within brands, • It’s necessary to know but they are sold with different codes the customer base • and understand how Only Prenatal has customers move its own private label from one brand to another
AIM OF THE PROJECT • Find customers that buys on different brands • Understand customers behaviors cross brand and cross channel • Define a new 1-to-1 campaign strategy • Move the customer from one brand to another during years • From childcare (0-11) to toys (0-99)
THE PROJECT SUPPORT THE TRANSFORMATION… The pillars to support transformation are: • Unified Product Catalogue • Customer Database to achieve the Single Customer View (per brand and cross brand) • Campaign Management to optimize the Engagement Process • Data Lake adoption to increase flexibility • Machine Learning to define a data-driven commercial strategy
3x THE PROJECT …WITH A BRAND NEW ARCHITECTURE Loyalty 4x Customer e-commerce Database Data Lake 3x 3x POS Product Catalogue Campaign Management
THE PROJECT DATA LAKE All this data are stored and harmonized inside the Data Lake Sell-out data : all the channels (stores, ecommerce sites) send their data to the lake Customer information : the customer inside the lake is unified, i also if he has multiple loyalty cards on different brands Product information : i is possible to unify all the product inside the lake to understand how the same product was sold in different brand stores
THE PROJECT DATA LAKE – A BIG ENABLER A Big Data centered architecture allows to: Add and remove brands in an easy way Define new cross-brand analysis Define new cross-brand marketing policies Add new data of other department (e.g. Logistic) to improve different processes
THE PROJECT MACHINE LEARNING – THE QUESTIONS TO ANSWER How many children does my customer have? Most of families declare only the first new born How old are the children? Which sex? What is the purchasing potential of my customer? + children +spending Am I fully exploiting the customer potential? Use MANY What products is my customer interested into? to understand ONE
3x THE PROJECT MACHINE LEARNING TO SUPPORT CAMPAIGN STRATEGY Loyalty 4x Customer e-commerce Database Data Lake 3x 3x POS Product Catalogue Campaign Management
THE PROJECT MACHINE LEARNING – USE CASE ROADMAP How many Purchasing Probability children does Curves Estimation my customer have? Child Age Estimation + Hidden Children Detection What is the purchasing potential of my Attribution Model customer? Product2Child Am I fully How old are Customer Lifetime Value exploiting the children? the customer Which sex? potential? Value Change Detection What products is my customer interested into? Product Recommender
THE PROJECT HIDDEN CHILDREN DETECTION & CLTV Net Past Customer Present Value Value Value Child 1 Child 1 Child 1 CLTV CL TV Net Past Customer Present Value Value Value Child 2 Child 2 Child 2
THE PROJECT CLVT - ACTIONABILITY CLTV CLTV «Unfreeze» customers Evaluate Marketing with high potential budget to invest in the customer Detect drops in spending behavior
THE PROJECT PRODUCT RECOMMENDER - ACTIONABILITY RECOM RECOMMEN ENDER DER ENGI ENGINE NE Checkout Coupons (fidelization) Website live suggestions (up-selling) Customized DEM (cross-selling)
THE PROJECT FINAL SUMMARY Understand as accurately as possible the number of children the customer has, their age and sex Understand the customer purchasing potential and calculate the CLTV Understand the customer tastes and recommend the right product at the right time Use the algorithms output as input for the Campaign Manager Personalized campaigns – Real time actions – Optimize retention
KEY LEARNINGS & TAKEAWAYS
Recommend
More recommend
