Statistics as Decision, Business and Marketing Science in the Disruptive Industry

A gentle introduction on how Statistics could confront the disruptive industry nowadays by acting as a decision science, business science, and marketing science. This session was presented at School of Business Management Institut Teknologi Bandung (SBM ITB) on a Guest Lecturer session.

Transcript

1. Guest Lecture Eps. Statistics as Decision, Business and Marketing Science in the Disruptive Industry Online Class @Hangout Google 21th of April, 2020. SBM ITB - Statistics Course Hello SBM ITB!
2. greetings! just call me Fiqry (without additional prefix/suffix)!!! I currently working as a Data Scientist @Bukalapak, also had been working as Technical Content Reviewer @Packt Publishing (working remotely) I also passionate on Time Series Analytics, Immersive Computing (VR & AR), and Gamification Business. That’s it ya!
3. disclaimer
4. “Statistics is the grammar of science.” —Karl-pearson
5. Statistics is the grammar of science Table of contents An end-to-end simulation, from deciding business problem to provisioning the solution High-level discussion: How data could decipher the business behavior today confronting the disruptive industry Get more detail: Statistics as Business, Decision, and Marketing Science Dealing the real business disruption 01 02 03
6. How data could decipher the business behavior today Confronting The disruptive industry 01
7. confronting the disruptive industry “Do you feel that nowadays our industries are moving too fast, going rapidly through beyond our imagination? Retail Transport Education Healthcare
8. we are in the middle of Age of Disruption
9. Disruptive Innovation principal Time Relatively quick for adaptation and adoption Quality Compel quality assurance, high standard service and significant impact Quantity Supermassive service range, huge user distribution and affordable cost
10. Disruptive Impact: ACT As Market Dominator OJEK PANGKALAN VS “Market-share Fairness or Consumer Fairness?” Ride-Hailing
11. Disruptive Impact: ACT As Market Dominator E-commerce
12. Disruptive Impact: ACT As Market Dominator E-commerce Smartphone Search Engine Social Media
13. Disruptive Impact: Companies are in a war, war of data “Data is a promising source to generate business insights, market prediction, and different strategies of marketing stuff while, Statistics is the scientific tools to make it happen” *Graph Rank: The Most Valuable Brands in 2020
14. Disruptive Impact: Data to Decipher business Behavior Problem solving, how to deal with the same problem but with a different way (solution) Automation concept, reducing operational cost and enlarging production scope Attract consumer, retain the existing one, and extent benefits personify Simplify intensify Decision Science Business Science Marketing Science
15. Text suggestions on Google Mail (Gmail) Google AI Services Data as the backbone of many AI services in Google Simplify
16. Amazon Warehouse management Streaming data as the fuel of robotics automation on warehouse management system IoT Implementation on Amazon Warehouse Intensify
17. Google AI Services Immersive Experience on Google Maps Data as the backbone of many AI services in Google Personify
18. And Many More data-driven services... Customer Segmentation Clusterize customer within their certain groups of habits Product Experimentation Discover product feature performance Market ForeCasting Predict the future of market wellbeings Recommender System Suggest customer by their historical activities
19. Confronting with disruptive industry meaning that dealing with Data & statistics
20. Statistics as Decision, Business, and Marketing Science Statistics is the grammar of science 02
21. Statistics is the grammar of science Getting more details about Statistics as a tools to distill data insights
22. Statistics is the grammar of science DeCision Science Marketing Science How to make a powerful business decision, by having low risk and gain immense impacts both qual and quant Business Science How to make a worthwhile business campaign, spend less money to get huge revenue How to accurately market a business product, with minimum cost use to attract potential loyal user
23. Statistics As Decision Science
24. Statistics as Decision Science Product Monitoring Dashboard Analytics, store main OKR/KPI metrics. Monitoring and Alerting system. Market Prediction Apply a time-series forecasting to predict the future market behavior. Product Experimentation Conduct a feature experimentation to test which alternative is better. Insights Models Insights
25. Statistics as Decision Science Product Monitoring Dashboard Analytics, store main OKR/KPI metrics. Monitoring and Alerting system. Insights Dashboard by Descriptive Statistics (Mean, Median, Mode, Percentile, Box-plot, etc.) Alerting by Confidence Interval
26. Statistics as Decision Science Data Forecasting by Time Series Analysis (Univariate/Multivariate) and Regression Analysis (Single, Multiple, Weighted, etc.) Market Prediction Apply a time-series forecasting to predict the future market behavior. Models
27. Statistics as Decision Science Experimenting by Pair Test (t-Test, Chi-Square,ANOVA, etc.) In the industry, this test also known as AB Testing Product Experimentation Conduct a feature experimentation to test which alternative is better. Insights
28. Statistics As Business Science
29. Statistics as Business Science Fraud Detection Cluster user activities, suspect user behavior that potentially harm the business life Churn Prediction Predict users by their histories, retarget them with special deals and campaigns Social Media Analytics Aware to public trends, shape a creative way to set sail for a new campaign Models Models insights
30. Statistics as Business Science Fraud Model by Multivariate Statistics (Cluster Analysis, Multidimensional Scaling, etc.) and Network Analysis Fraud Detection Cluster user activities, suspect user behavior that potentially harm the business life Models
31. Statistics as Business Science Churn Prediction by Regression Analysis (Multiple, Logistic) and Supervised Machine Learning (Random Forest, XGBoost, etc.) Churn Prediction Predict users by their histories, retarget them with special deals and campaigns Models
32. Statistics as Business Science Social Media Analysis by Descriptive Statistics (Mean, Median, Mode, Percentile, Box-plot, etc.) and Supervised Machine Learning (Topic Clustering, Sentiment Analysis) Social Media Analytics Aware to public trends, shape a creative way to set sail for a new campaign insights
33. Statistics As Marketing Science
34. Statistics as Marketing Science Customer Segmentation Group users into segments, run with different type of treatments Recommender System Read personal activity of users, suggest them with some similar interests Product Ads Show research results carefully to public, convince public with the right data insights Models insights
35. Statistics as Marketing Science Segmentation by Multivariate Statistics (Cluster Analysis, Multidimensional Scaling, etc.). Such as LRFM models (Length, Recency, Frequency, and Monetary) Customer Segmentation Group users into segments, run with different type of treatments insights
36. Statistics as Marketing Science Recommendation Cluster by Matrix Factorization (Content-Based Filtering, Collaborative Filtering) and Multivariate Statistics (kNN segmentation, Multidimensional Scaling, etc.) Recommender System Read personal activity of users, suggest them with some similar interests Models
37. Statistics as Marketing Science Recommender System Read personal activity of users, suggest them with some similar interests Models
38. Statistics as Marketing Science Research is conducted by Descriptive Statistics (Proportion) and Inferential Statistics (t-Test, Chi-Square Test, ANOVA, etc) Product Ads Show research results carefully to public, convince public with the right data insights
39. An end-to-end simulation, from deciding business problem to provisioning the solution Dealing the real business disruption 03
40. Dealing the real business disruption A compelling story in tech industry, how do they create disruptive innovation
41. The story behind Tech Product Development Flow Tribe/Squad-level Discussion Technical level, talk about agile sprints, engineering readiness, product design, user research needs, and data input. Product-Level Discussion Management level, talk about the product and business continuities Output 1. OKR/KPI to achieve 2. Feature to make/enhance/remove 3. Set target deadline 4. Reporting Member 1. C-level (COO) 2. VP-level (VP of Product, Design, Data) 3. Tribe Lead (Business Team) 4. Product Manager (Related feature) Output 1. Sprint Plan, Preprod 2. User Research 3. AB Testing 4. Deployment to Production Member 1. Product Managers 2. Data (Analyst, Scientist) 3. Software Engineers (BE, FE, SET) 4. Quality Assurance (QA) 5. UI Designer 6. UX Researcher Case: Homepage Revamp for New Users Ideation Reporting
42. The story behind Tech Product Development Flow Business Problem Discovery (OKR/KPI/North star metrics) Product Feature Ideation Data Quant & Qual Insights New users tend to idle on homepage. Interview result shows that idle users are confused to make transaction Business Meeting Discussion between PM, TL, and VP-level (or C-level). What feature will be built to solve problem Product Development Building feature Software Engineers, UI Designers create the proposed feature (Proof of Concept, Raw Design) Product Experimentation AB Testing Separate 50:50 users, group A will be exposed. Group B will not be exposed Product Release Full Deployment Deploy 100% to the system, so 100% users will be exposed the winning result from AB Testing Product Report Data Reporting Create dashboard for the new feature, also make the alerting system. This will be used to monitor the new feature data
43. The story behind Tech Product Development Flow Product Experimentation AB Testing Separate 50:50 users, group A will be exposed. Group B will not be exposed Segment A Segment B Variant P-value Variant Win-prob View 135 170 0.252 74.8% Click 89 90 0.023 97.7% Transact 80 45 0.008 99.2% Segment A Segment B CVR (View to Click) 78/135 = 66% 90/170 = 52% CTR (Click to Trx) 80/89 = 89% 45/90 = 50% User id = 123456 Segment = A Total View = 9 Total Click = 4 Total Transact = 3 Data Per User AB Test Metrics
44. Conclusion, lesson learn Key Takeaways 04
45. Talk Key Takeaways! Statistics as Decision, Business and Marketing Science Statistics is the grammar of science Jargons to confront the disruptive industry: Simplify, Intensify, and Personify Confronting the disruptive industry
46. “I don't believe in the glory and the dream. I believe in statistics.” —Amy Gentry
47. www.menti.com 67 87 40
48. Hatur Nuhun! That’s all folks, hope you enjoy! Having disruptions? Just ask here fiqryrevadiansyah@gmail.com Telegram: @fiqryr WA: +62 857 2031 6671 linkedin.com/in/fiqryrevadiansyah
49. Credits Special thanks goes to: ◂ Presentation template by Slidesgo ◂ Icons by Flaticon ◂ Images & infographics by Freepik ◂ Author introduction slide photo created by Freepik ◂ Text & Image slide photo created by Freepik.com ◂ Some videos taken from Youtube.com ◂ https://wallpaperaccess.com/purple-space ◂ https://blog.weekdone.com/understanding-okrs-better/ ◂ https://smac-group.github.io/ts/basic-elements-of-time-series.html ◂ https://www.optimizely.com/optimization-glossary/ab-testing/ ◂ https://twitter.com/EU_Taxud/status/1128611157491712001/photo/1 ◂ https://www.analyticsvidhya.com/blog/2019/05/practical-introduction-prescriptive-analytics/customer-chur n-edit/ ◂ https://tweetreach.com/twitter-analytics-report/ ◂ https://sebastianraschka.com/Articles/2014_twitter_wordcloud.html ◂ https://www.retailreco.com/blog/rfm-analysis-for-customer-segmentation-in-ecommerce/