From Data to Data Driven - Applications that will change your business
Download as PPTX, PDF8 likes2,348 views
The document discusses how businesses are moving from simply analyzing large datasets to developing data-driven applications that can actively guide processes in real-time based on insights gleaned from data. It notes that while platforms like Hadoop enable cost-effective storage and processing of large, diverse datasets, true value comes from building applications that deliver insights to users and influence outcomes. The document advocates an approach combining traditional data warehousing with real-time analytics to power applications that learn continuously and provide smarter, personalized experiences over time.
From Data to Data Driven - Applications that will change your business
- 1. From Data to Data Driven Applications That Will Change Your Business With Matt Aslett/451 Research and Steven Noels/NGDATA
- 2. Š 2014 by The 451 Group. All rights reserved Big Data â Virtuous Circle Increased use of interactive applications and data-generating machines New commercial opportunities for analyzing previously ignored data Increased desire to store and process all available data New infrastructure investments to support new data processing software More economically feasible to store and process previously ignored data
- 3. ď§ Important step on the path to influencing behavior via predictive analytics ď§ For example ad/content serving, fraud detection, risk profiling, and customer retention Š 2014 by The 451 Group. All rights reserved Growing Analytic Ambitions REPORTING - What happened ANALYSIS - Why did it happen? PREDICTIVE - What will happen? OPERATIONAL - What is happening now? INFLUENCIAL - How to influence what happens next? PREDICTIVE ANALYTICS MACHINE LEARNING OPERATIONAL INTELLIGENCE
- 4. PREDICTIVE - What will happen? ď§ Data scientists ď§ Historical data ď§ Automating processes ď§ Data-driven people Š 2014 by The 451 Group. All rights reserved Different Usage Patterns REPORTING - What happened ANALYSIS - Why did it happen? ď§ Business analysts ď§ Historical data ď§ Manual processes ď§ People-driven dashboards OPERATIONAL - What is happening now? ď§ Business analysts/data scientists ď§ Real-time data ď§ Data-driven dashboards INFLUENCIAL - How to influence what happens next? ď§ Users/customers ď§ Automated processes ď§ Historical and real-time data ď§ Data-driven applications
- 5. Expanding Data Platforms ď§ The desire to take advantage of that opportunity for competitive advantage led many enterprises to expand their data processing platforms ď§ Taking advantage of Hadoop and MapReduce to enable the processing and analysis of multi-structured APPLICATIONS OPERATIONAL DATABASE DATA WAREHOUSE Š 2013 by The 451 Group. All rights reserved data ď§ Where do we go from there? ď§ How to maximize the potential investment in new platforms and move from reactive to proactive analytics and data-driven applications? STRUCTURED DATA STRUCTURED DATA AD HOC ANALYTICS PRE-DEFINED REPORTING PREDICTIVE ANALYTICS HADOOP MULTI-STRUCTURED DATA
- 6. Hadoop is Not a Panacea ď§ It does enable the economic storage and processing of data ill-suited to the cost and functional capabilities of traditional data management technologies ď§ Hadoop requires a sophisticated skill set that was not prevalent within many organizations Š 2013 by The 451 Group. All rights reserved ď§ MapReduce programmers, HDFS administrators ď§ The skillset has become more prevalent, and Hadoop itself easier to use ď§ Support for existing skills and tools (e.g. SQL analytics) ď§ However, most initial projects were engineering led and limited to a discreet set of users ď§ âData scientistsâ, specific departments/use-cases (e.g. search) ď§ Net result: Hadoop offers a huge opportunity to store and process data that was previously ignored, but the true potential is not being realized
- 7. HIVE PIG HBASE STORM GIRAPH MAHOUT Š 2013 by The 451 Group. All rights reserved Hadoopâs Potential is Built on Flexibility ď§ Hadoop can be used as a platform for: ď§ Reporting, Ad-hoc analytics ď§ Statistical analytics, Predictive analytics, Machine learning ď§ Graph and stream-based analytics ď§ Initially, all programmed via the MapReduce batch process MAPREDUCE HDFS AD HOC ANALYTICS PRE-DEFINED REPORTING PREDICTIVE ANALYTICS MACHINE LEARNING
- 8. ITERATIVE ANALYTICS HIVE PIG TEZ (optional) Š 2013 by The 451 Group. All rights reserved Hadoopâs Potential is Built on Flexibility ď§ Next-generation Hadoop, with YARN and Spark (in-memory) enable: ď§ Multiple services to be run on data stored in HDFS, simultaneously ď§ By-passing the MapReduce batch process, enabling interactivity HDFS YARN HBASE SQL-on- HADOOP, STORM, GIRAPH, OTHER MAPREDUCE SPARK TACHYON (developing) SHARK MLLIB (developing) GRAPHX (developing) SPARK STREAMING SPARKR, SPORK, OTHER AD HOC ANALYTICS PRE-DEFINED REPORTING PREDICTIVE ANALYTICS MACHINE LEARNING MAHOUT OPERATIONAL INTELLIGENCE
- 9. Applications are Key to Fulfilling that Potential HIVE PIG HDFS TEZ (optional) Š 2013 by The 451 Group. All rights reserved YARN HBASE SQL-on- HADOOP, STORM, GIRAPH, OTHER SHARK MAPREDUCE MLLIB (developing) GRAPHX (developing) SPARK STREAMING SPARKR, SPORK, OTHER MAHOUT APPLICATIONS AD HOC ANALYTICS PRE-DEFINED REPORTING PREDICTIVE ANALYTICS MACHINE LEARNING ITERATIVE ANALYTICS OPERATIONAL INTELLIGENCE SPARK TACHYON (developing)
- 10. Š 2013 by The 451 Group. All rights reserved Putting it All Together ď§ Despite all the focus on data platforms, it is the applications that deliver the value ď§ To the business ď§ To the user ď§ And generate the real-time data, to be combined with historical data and new approaches to data processing to deliver data-driven ď§ Delivering that requires not just Hadoop ď§ Real-time data processing ď§ Predictive analytics ď§ Machine learning ď§ Recommendation engine ď§ But the underlying technology needs to serve not just business analysts and data scientists but data-driven applications that drive automated business processes
- 11. From Data to Data-Driven Delivering Value to the Business
- 12. Defining Data-Driven Applications Data-driven applications act on any volume of real-time data to actively guide and optimize business processes based on continuously growing insights on customer behavior, intent and preferences. Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 12
- 13. Data-Driven Applications Allow You to⌠Proactively Steer Processes Go Beyond Insights Have a Closed Feedback Loop Benefit from Continuous Learning Get Smarter Over Time Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 13
- 14. Applying Data-Driven to the Customer Experience Relevance? Awareness? Value? Timing? Clarity? Customer insight is limited to a subset of available data - Limiting relevance and timeliness of offers Social Data Company & Customer Activity Customer Web & Mobile Website & online apps Mobile App Server Customer Channel Campaigns Mail SMS Print Broadcast Customer Service Desk Customer CRM Systems Offers direct mail ATM web Agent, IVR email mobile chat Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 14
- 15. Data-Driven â More than Big Data Graphs⌠Enterprises are NOT looking for new presentations of their data anymore, but for RESULTS and for solutions and people to achieve them Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 15
- 16. Data-Driven â Answering the Tougher Questions⌠Which top hundred customers are likely to buy my product X today? I have a customer - what are the top 3 products he is likely to buy? What is the best channel to connect with my customer, and when? How can I turn around my most valuable potential churners? Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 16
- 17. Pre-requisites to Deliver Data-Driven Customer Experiences Preferences Context Affinities Behavior Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 17
- 18. DNA Towards a New Architecture App Phone Mobile www STB Network Shop Call Centre CDR POS Personalization Acquisition Journey ARPU / CLTV Marketing Advertising Churn Fraud App Phone Mobile www STB Network Shop Call Centre POS CDR Personalization Acquisition Journey ARPU / CLTV Marketing Advertising Churn Fraud Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 18
- 19. Lily Real Time Personalization Within the Current Enterprise Architecture â Improving existing BI landscape Operational Systems Contract/Product Data Customer Operational Data Data integration ETL or Lily API based Batch Lily Enterprise Customer Interaction Data Campaign Data ERP/CRM Data Data Warehouse Data External Data Reference Data Reporting / Analytics Enterprise BI and reporting Enterprise Analytics Applications Marketing and Social Data Web and Mobile Customer Website & Social Mobile Apps Channel / Campaigns Mail SMS Print Broadcast Sales Office Agent / Advisor Service Desk Marketing Campaign Mgt Customer CRM and IVR Systems Integration Real Time Integration Lily Connector Based Feedback Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 19
- 20. Big Data Real Time Analytics What is Different? Batch Systems Real-Time Systems Traditional Analytic Systems What will happen? What do I need to do? What has happened? Why? Large Volumes Multi-structured Data Sampled Data Structured Data âMy traditional BI environment will give the answers tomorrow of yesterdayâs problemâ - CIO Fortune 50 Bank Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 20
- 21. Applied to Subscription-based Businesses Customer Lifetime Value (Inbound, Outbound, Risk) ⢠Personalization ⢠Offer recommendations based on individual preferences ⢠Customer Service assistance ⢠Manage risk, fraud prevention, Reduce customer churn ⢠Personalization, Customer Services ⢠Better targeting, micro campaigns, social leaders ⢠Reduce customer churn, price sensitivity ⢠Media and IPTV, audience measurement ⢠Personalization, Cross sell/upsell ⢠Recommendations â ads, content, etc. ⢠Audience measurement ⢠Customer Services Banks Telecoms Media and Publishers Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 21
- 22. From Data to Data Driven Applications Listen Bigger. VOLUME Learn Faster. Volumes of Data Availability Questions Answered SPEED Execute Smarter. ANSWERS Unknown Unknowns Known Unknowns DW/BI Zettabytes Exaabytes Petabytes Terabytes Gigabytes DW/BI Seconds Minutes Hours Days Weeks DW/BI Known Knowns Copyright 2014 NGDATAÂŽ, Inc. Confidential â Distribution prohibited without permission 22
- 23. Questions?
- 24. Go from Data to Data Driven Applications To learn more about: NGDATA and Lily go to NGDATA.com or info@ngdata.com 451 Research â 451Research.com or matthew.aslett@451research.com