ARTIFICIAL INTELLIGENCE & INNOVATION
SUMMARY 1. STRIVE FOR OPERATIONS EXCELLENCE THROUGH AI 2. INNOVATION AI & INNOVATION 2
SUMMARY 1. STRIVE FOR OPERATIONS EXCELLENCE THROUGH AI 2. INNOVATION AI & INNOVATION 3
AI AND INNOVATION SIMPLIFIED – ILLUSTRATIVE DATA SCIENCE KERING AND ARTIFICIAL INNOVATION INTELLIGENCE TEAM'S FOCUS Kering AI factory's focus Image Recognition AR & VR Chatbot Assistants Industrialization Blockchain Disruptive materials Delivery 4.0 Voice Commerce 5G Adoption Proof of concept Awareness TECHNOLOGY MATURITY Sources: GrandViewResearch; BI Intelligence; Statista; Global Data; eMarketer AI & INNOVATION 4
AI IS BRINGING A WIDE SCOPE OF GAME-CHANGING APPLICATIONS PLANNING Topic extraction Deep learning & OPTIMIZATION Translation Classification Predictive analytics LANGUAGE MACHINE PROCESSING LEARNING Machine vision EXPERT VISION SYSTEMS Image recognition Speech to text Text to speech SPEECH ROBOTICS Used for current projects Being assessed for future projects AI & INNOVATION 5
THE AI FACTORY RELIES ON A LARGE RANGE OF DATA SOURCES SALES DATA CRM DATA Sales in value Client contact Product volumes Segmentation Sales location and date Purchase behavior Opt-in/Opt-out PRODUCT DATA WEBSITE DATA Price Web views Style Color & Size Online Conversions Product Attributes SUPPLY DATA OPEN DATA Inventory in network Postal code Geolocation mapping DATA LAKE Replenishment data Special events Store data & AI PLATFORM AI & INNOVATION 6
AI WILL BRING A COMPETITIVE ADVANTAGE TO BRANDS: HENCE, WE ARE LEVERAGING THE BEST CAPABILITIES AND TALENTS TO ACCELERATE 2018 2019 2020 Phase 1 Phase 2 Phase 3 Prototyping & validating opportunities Building capabilities Scaling up & industrializing Implementing opportunities BUILD A UNIQUE TEAM IMPLEMENT CUTTING FOCUS ON MVP ENSURE CLOSE OF TALENTS EDGE TECHNICAL AND INDUSTRIALIZATION INCLUSION OF BUSINESS CAPABILITIES RATHER THAN POC REQUIREMENTS • • • • One team / One roof / Upgrading Kering data lake Strategic bias in favor Onboarding the brands from One floor blending data to store all data sources of Minimum Viable Products day 1 scientists, data engineers, and ensure data availability; (= the most minimal form • Starting all projects data developers Manage huge amount of a complete solution) to test with a sponsor brand & data managers of unstructured raw data in real conditions as soon • “The hardest part of AI as possible • • Leverage agile method to Leverage cloud platform is not the code, it is the • foster innovation and adopt (scalability, AI) and ability to Aiming for rapid change management around” new technologies rollout models in production industrialization at full scale for projects with proven added value AI & INNOVATION 7
WE PRIORITIZED AI PROJECTS AMONG A LARGE SPECTRUM ON THE VALUE CHAIN DESIGN BUYING PRODUCTION LOGISTICS STORE MARKETING SALES & MERCH. LAYOUT Buying Supply chain Trend Demand CRM & Markdown I P product P S Replenishment & Store-to-Store S C prediction planning clienteling optimization scoring optimization Automatic Store Price Precision P product I location I I Premium media tagging evaluation optimization Collection Store Ideal store I structure I I workforce layout optimization planning Customer- I centric store assortment I Performance management: Sales forecasting (global, country, store level) Prototyping Building Scaling up I P C S Ideation & validation capabilities & industrializing AI & INNOVATION 8
FOCUS ON SUPPLY CHAIN PYTHAGORAS PROJECT: ASSIST PLANNERS IN OPTIMIZING STORE REPLENISHMENT STRATEGIES LIVE IN JUNE 2019 EXPECTED BENEFITS • Assist planners in optimizing product quantities to ship to stores in order to reduce What? inventory shortages and overstocks INCREASED FORECAST ACCURACY • • ~20% more accurate on one Start with two categories in Europe of the categories • Forecast "Newness" products with no historical data • Develop a new AI-driven short-term sales forecast model and (as a second step) a • Maximize sales at full price replenishment-optimization model in close • Maximize gross margin How? relationship with Gucci replenishment teams • Integrate AI outputs in existing planning tools LOWER RISK OF INVENTORY SHORTAGES OR OVERSTOCK Thanks to an improved reaction to market Forecast Rollout of forecast variability, in particular for items with a few live in to all products and weeks of historical data planner’s tools regions launched Key SMOOTHER PROCESS milestones 06.19 07.19 08.19 09.19 10.19 11.19 12.19 01.20 Thanks to a more reliable forecast Launch study and simplified validation tool on replenishment optimization AI & INNOVATION 9
PYTHAGORAS RESULTS ARE VERY ENCOURAGING FORECASTS VS. REAL SALES COMPARISON PRODUCT CATEGORY A, EUROPE Pythagoras forecast Current forecast Real adjusted sales 2/12/18 9/12/18 13/1/19 27/1/19 10/2/19 17/2/19 24/2/19 10/3/19 17/3/19 24/3/19 31/3/19 18/11/18 25/11/18 16/12/18 23/12/18 20/1/19 3/2/19 3/3/19 IMPROVEMENT IN SALES FORECAST ACCURACY OVERALL PERIOD +20% AI & INNOVATION 10
FOCUS ON PRICING PROJECT: FIND THE OPTIMAL MARKDOWN LEVEL OBJECTIVE: LIVE IN H2 2019 EXPECTED BENEFITS • Find out the optimal markdown level for each product What? • Start with women’s and men’s shoes in Europe for new 2019 Fall season HIGHER SALES • Develop new AI-driven price-sensitive sales HIGHER GROSS MARGIN forecast model How? • Develop a pricing optimization model OPTIMIZE WORKING CAPITAL • Promising results for the sales forecast : high accuracy, although some outliers need to be understood First • Price sensitivity analysis : forecasts PROTECT BRAND EQUITY results & are dependent on prices and this sensitivity next steps is variable across products • We are running optimization methods to find the best combination of discounts AI & INNOVATION 11
BEYOND SUPPLY AND SALES, WE BELIEVE THAT AI WILL HAVE A WIDE IMPACT ON A LARGE RANGE OF KERING ACTIVITIES AI CAPABILITIES LEGAL CUSTOMER SERVICE • • Counterfeit detection Inbound message ranked by priority Planning • Trademark tagging & Optimization • Personalized e-mail on product images _ proposition Machine Learning _ Image Recognition _ Language Processing _ Machine Vision _ Expert Systems MEDIA TALENTS • • Media spend optimization Resume screening (precision marketing) • Employee churn prediction Used for current projects • Career path counselling Being assessed for future projects AI & INNOVATION 12
ARTIFICIAL INTELLIGENCE & INNOVATION 1. STRIVE FOR OPERATIONS EXCELLENCE THROUGH AI 2. INNOVATION AI & INNOVATION 13
INNOVATION AT KERING OUR MISSION • Have a 360° view of key innovation trends • Prioritize key disruptions • Drive value by integrating disruptions into our business (through proof-of-concept and investment) AI & INNOVATION 14
INNOVATION GOVERNANCE KERING INNOVATION ADVISORY BOARD Gucci Saint Laurent Alexander McQueen Balenciaga INNOVATION EXPLORERS W&J brands Kering Eyewear Bottega Veneta EXTERNAL NETWORK OF KEY PARTNERS AI & INNOVATION 15
INNOVATION ROADMAP 1 2 3 DISRUPTIONS TECHNOLOGIES ENABLERS IMPACTING OUR IMPROVING VALUE BUSINESS MODEL TO CUSTOMERS & OUR PERFORMANCE MATERIAL DISRUPTIONS IN-STORE / ONLINE / INNOVATION ECOSYSTEM CLIENT SERVICES SET-UP NEW BUSINESS MODELS In-store: Technologies to augment Partnership with Venture Capitalists Client Advisors New ways to consume luxury (e.g. subscription, Start-up scouting second hand, rental) Online: Technologies to improve user experience and drive more conversion New ways to engage with consumers TECH SCOUTING Client services: AI-based technologies to augment Client Services Advisors and improve Blockchain performance and monitoring Voice / Chatbot Image recognition INTRAPRENEURSHIP / CULTURE Idea crowdsourcing Co-design AI & INNOVATION 16
INNOVATION FUNNEL INVESTIGATION PROJECT LIVE EXPERIMENTATION 1 2 3 4 IDEAS POC MVP ROLLOUT • Test feasibility • Assess tech/solution • Roll-out at small scale • maturity Test fit with • Run in production FOCUS business need • Maximize learnings • Assess what is at stake • Deploy for the least efforts • Identify features that • Identify use case deliver the most value Prioritized list Go / No go on Industrialized RESULTS Feedback on solution of use cases feasibility & usability version of solution Note: POC: Proof of concept / MVP: Minimum Viable Product AI & INNOVATION 17
RETAIL VOICE ASSISTANT 1 2 3 4 IDEAS POC MVP ROLLOUT TOPIC Voice & Text assistant to help Client Advisors with retail procedures in order to maximize selling time on the sales floor BENEFITS FOR THE USER • Save time: Get an easy and immediate access to complex procedures • Better serve: Provide answers on after sales cases without leaving the customer • Better train: Facilitate the training of new hires APPROACH • Small-scale field experimentation to test in quick, agile way • Technology: Natural Language Processing • KPIs: Understanding / Accuracy / User experience • Next step: Deploy in additional stores and add new languages AI & INNOVATION 18
Recommend
More recommend
