Ad
More Related Content
What's hot (20)
Similar to Integrate Your Data Science & Omni-channel Strategy to Reduce Cost and Increase Optimization - Daniel Gremmell (20)
Ad
Recently uploaded (20)
Ad
Integrate Your Data Science & Omni-channel Strategy to Reduce Cost and Increase Optimization - Daniel Gremmell
- 1. Integrate Your Data Science & Omni- channel Strategy to Reduce Cost and Increase Optimization DATE: Nov 2019 Daniel Gremmell VP, Data Science
- 2. 2 This Session Will Seek to Cover: Build a data science strategy that solves problems along your customer journey as well as your value chain. It can be overwhelming understanding where to start with data science. Understand how and where to solve problems in a digital and physical setting. Understand when to use human in the loop machine learning to optimize creative flows. Creative processes are naturally challenging and contain large amounts of insight that is hard to qualify. Understand how we have implemented machine learning in our culinary process. Integrate your business strategy with your data science strategy for maximum value. A data science strategy is awesome, but only effectual if it is integrated with a business strategy. Understand how we have integrated our data science and business strategies.
- 3. 3 Section 1 About Plated and Albertsons
- 4. 4 About Plated and Albertsons Plated is an Albertson’s owned brand that delivers a meal experience to our customers through an Ecommerce and Retail channel. Plated was originally a subscription service and has branched into omnichannel with the Albertson’s acquisition. We work hard to develop a lifestyle solution that meets the needs of our customers and solves their daily problem of what they should have for each meal in a convenient way. Check us out at Plated.com for more information and check out our blog!
- 5. 5 Plated Data Science Problems There are numerous applications and problems the data science team at Plated has solved such as: ○ Weekly recommendations of dinners ○ Forecasting demand with little to no history ○ Blog content recommendations ○ Using simulated annealing to build a menu ○ Multiobjective add on product recommendations ○ Identifying Albertson’s Customers that look like Plated customers. ○ Using favorited recipes to recommend similar recipes. ○ Using Albertson’s basket data to recommend a Plated meal. ○ Predicting Plated Customers likely to churn.
- 6. 6 Section 1 Data Science Strategy
- 7. Data Science Maturity Model* Business Monitoring Business Insights Business Optimization Business Metamorphosis Bill Schmarzo - “The Big Data MBA”
- 8. Prescriptive Analytics Strategy 8 Prescriptive Analytics Consumer Value Generation Strategy Value Chain Strategy Operations / Internal Processes Marketing Algorithms Personalization Data Products
- 9. Consumer Value Generation Consumer value generation breaks down into two application areas: 1. Personalization - crafting the customer experience to suit their traits and behaviour. Examples are product recommendations and content recommendations. 1. Data Products - An actual product driven by data science and machine learning. Example is Look n’ Cook product feature. 9
- 10. 10 Value Chain Strategy The value chain is defined as the set of activities we must accomplish to deliver our product to our customers. It looks at the activities as a system with multiple subsystems. It includes: 1. Recipe Development 2. Menu Construction 3. Sourcing 4. Forecasting / Planning 5. Distribution / Logistics 6. Many Others
- 11. 11 The Data Science Value Chain Strategy Machines…. The Plated / ABS Data Science Value Chain Strategy consists of: ○ Prescriptive machines firing at points in the value chain where optimization and decision making is required. ○ Provides efficiency by freeing up humans to progress the business and allows for optimal decisions at all times. ○ Example of this is the integration of our recommendations and forecasting engines.
- 12. 12 Section 2 Implementing Data Science in Creative Processes
- 13. RECIPE CONTENT At Plated, Recipe Generation is a very artistic process. However, we have infused our process with machine learning and are expanding further into human in the loop solutions.
- 15. 15 Section 3 Integrating Data Science and the Omni-Channel Strategy
- 16. 16 Data Science and Strategy One of the most important things about implementing data science is it must become part of the culture. ○ Data Leader should sit on the leadership team and participate in strategy planning. ○ The data science strategy should enhance and flow back to the overall company strategy. ○ Data science maturity is a long term goal that changes focus as the company progresses in the journey. ○ A successful omni-channel strategy revolves around building an immersive experience that transcends retail and online.
- 17. 17 Strategy is about making choices. To win a company must do some things and not others.* *Lafley and Martin - “Playing to Win”
- 18. 18 The Essence of Our Strat Planning Process Every Strategy is Uncertain The whole idea of our process is to: ○ Brainstorm multiple creative and distinct ways to win with our customers. ○ Collect points and assumptions of what must be true and highlight where the most uncertainty lies. ○ Design quick tests to validate or invalidate assumptions around where uncertainty lies. ○ This avoids wide but shallow data analysis by focusing on narrow but deep questions to test.
- 19. 19
- 20. 20
- 21. 21 DS cascade works with the same process and is meant to support the higher level company strategy.
- 22. 22 Omni-channel Data Science Application Connecting retail customers with those who look like E- Commerce customers to build the omni-channel footprint. Used to target retail customers with online promotions, recommendations and products to build a pure omni-channel experience. Predicting Omni-Channel Customers With the mix of Albertson’s retail and E- Commerce data, we were able to: ○ Find overlap of existing Plated/ABS E- commerce and Retail Customers. ○ Extract known positive cases along with unlabeled data. ○ Fit multiple models to classify known cases and unlabeled cases that look like positive known cases. ○ This allows us to leverage ABS data such as marketing segments, attributes and transactional data and market an omni- channel experience directly to these people.
- 23. 23 A Few Omni-channel Tips ● You cannot be all things to all people. Make choices. ● It is not online or retail, it is both. ● Use a digital experience you can measure and optimize to immerse retail shoppers. ● Change your company’s measurement systems that consider channels in isolation. ● Use machine learning and statistical attribution models to build profiles of your consumers to connect retail with online. ● Can’t make the connection? Build an experience that encourages sharing and digital interaction..
- 24. T H A N K Y O U !