Your SlideShare is downloading. ×

Mastering Big Data: The Next Big Leap for Master Data Management


Published on

By leveraging both big data and master data, organizations can provide a true 360-degree view of critical master data entities and produce actionable insights gleaned from big data sources.

By leveraging both big data and master data, organizations can provide a true 360-degree view of critical master data entities and produce actionable insights gleaned from big data sources.

  • Be the first to comment

No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

No notes for slide


  • 1. Mastering Big Data: The Next BigLeap for Master Data ManagementBy leveraging big data, organizations can provide a true360-degree view of critical master data entities; meanwhile,master data can help convert information gleaned from bigdata sources into actionable insight.Master data management (MDM) is critical tothe success of many big data analytics projects.Why? Because it encompasses people, processesand technologies that ensure the accuracy,uniformity and uniqueness of data entities sharedby multiple business units within an organiza-tion. (See our white paper “Innovations in MDMImplementation: Success via a Boxed Approach”for additional insights.) Big data, on the otherhand, refers to the large quantities of human-or machine-generated bits and bytes that aregenerated at high speeds from multiple sources,including not only posts and comments on socialmedia but also images, videos, PDFs, RFID data,point of sale data, call center comments, datagenerated by enterprise systems and more. A bigdata analytics strategy involves collecting thisdata in order to create meaningful information.All this data, in various shapes, sizes and forms,represents valuable information that needs tobe standardized and stored in a meaningfulway within a central repository accessible by allbusiness units. MDM provides a trusted, centralpoint of storage and access to this information.This whitepaper describes how master datamanagement and big data analysis need to workhand-in-hand to provide a 360-degree view ofan organization’s critical data entities. (Whilethis paper is largely focused on social data andcustomer MDM, there are many other applica-tions of big data not explored here.)The Relationship BetweenBig Data and MDMMDM provides a meaningful context to big dataand a panoramic view of critical data entities,such as product, customer, employee, agent,location, etc. (see Figure 1, next page). With all thisinformation stored in a central trusted repository,all business units have access to a single versionof the truth, on demand. In the absence of such amaster data hub, records are typically stored insilos, such as loyalty systems, CRM systems, POSsystems, Web portals and big data repositories.Such systems usually have their own version ofthe truth, causing headaches that ripple acrossthe organization, from substandard customerservice and bad business intelligence, throughimpaired operational efficiency (see Figure 2,next page).While MDM is crucial to the success of many bigdata projects, it is important to note that not allbig data initiatives require an MDM solution. MDMis needed only for big data initiatives that involve• Cognizant 20-20 Insightscognizant 20-20 insights | may 2013
  • 2. 2master data entities, such as customer, employee,agent, product, location, etc.The Scope of Big Data and MDMWhen data is extracted from most big datasources, it contains two parts. The first part isentity attribute-level data, or data that can beused to identify an entity. The other is transac-tion-level data that is generated by each entity.For instance, if the big data source is Facebook,the two types of data include:• Attribute-level data: Facebook ID, phonenumber, name, address, etc.• Transaction-level data: Conversations,comments, likes, etc.The big data extraction layer will pass the attri-bute-level data to the MDM system and thetransaction-level data to the big data hub, tobe analyzed by a big data analysis tool. Afteranalysis, the brand advocate score, purchaseintent, etc. will be passed to the MDM system tocreate the 360-degree view of the customer.To help define the scope of big data, consider thefollowing:• MDM will act as a repository of the followinginformation types obtained from big datasources:>> Attribute-level data.>> Information derived from transaction-leveldata.cognizant 20-20 insightsFigure 1A 360-Degree View of Jane DoeFigure 2With or Without MDM?Customer storepurchase informationand the informationentered in loyalty card.Basic customer information,campaign information,targeted mailer, couponinginformation.Social conversations,whether the customer isa net promoter, on theverge of switching,network reach, etc.Web purchase history,information enteredduring checkout.Central reliablerepositoryCustomer store purchaseinformation and theinformation entered inloyalty card.Basic customerinformation, campaigninformation, targetedmailer, couponinginformation.Social conversations,whether the customer is anet promoter, on theverge of switching,network reach, etc.Web purchase history,information enteredduring checkout.Call center conversation,number of problemsresolved, dissatisfiedcustomers.No reliable,360-degree viewCRM SystemCRM SystemSocial Data – Big DataSocial Data – Big DataWeb Portal Web PortalIVR Portal IVR PortalPOS System POS SystemCall center conversation,number of problemsresolved, dissatisfiedcustomers.Unstructured Internal DataUnstructured External DataThird-partyInformationNominalInformation• Call center conversations• Point of sale data• Conversations• Reviews• Blogs• Images• Videos• Credit score• Professional affiliations• MembershipsName: Jane DoeAddress: 1 Acme Road,Kiwi NJ 00000Phone: 555-555-5555Age: 23
  • 3. • MDM will cleanse, standardize and de-duplicatethis entity-level data and maintain audit trailsand hierarchies.• MDM will not store entity-level transactionaldata, such as conversations, transactions, etc.• MDM will not analyze big data.In the absence of an MDM system, both attribute-level and transaction-level data can be stored inbig data hubs. But such data stores are meant forheterogeneous, unstructured or semi-structureddata that is intended for analytics purposes;hence, they are not optimal for managing sharedenterprise master data entities. Additionally:• Data models in big data hubs are optimizedfor analytics, while MDM data models areoptimized for entity resolution.• A big data hub stores all information gatheredfrom a big data source; however, these recordsare neither unique nor cleansed and rectified.Hence, finding an authoritative golden copycan be difficult and time-consuming.• The sheer volume of data in the big data hubwill not allow a quick search for entity-levelattributes for customer service or decision-making purposes.• Big data hubs are built for a specific purpose— big data analysis. Hence, strong entitlementand hierarchy management are not included.MDM for Big Data AnalyticsTo handle the demands of big data, next-genera-tion MDM systems must successfully handle thevolume, speed and formats across all data types.Required fundamental changes to MDM include:• A data model to store big data-specificattributes, such as social media IDs, opt-ins,brand advocate scores, social network hierar-chies, product preferences, purchase intent,churn possibility, etc. As previously mentioned,some of these attributes will be fed into theMDM system from big data analytics systems;others will be fed directly from the source.• Improved MDM integration architectures tohandle the speed at which this data is bothadded and updated.• Advanced match and merge processes,especially for social analytics, which is amongthe most widely applied use cases in big data.The biggest challenge with social analytics issuccessfully identifying the social profiles ofan organization’s existing customers. This isespecially true since multiple results can bereturned when searching for common names.Also, many people use aliases rather than realnames in their social profiles, making it evenmore difficult to identify these profiles.• Advanced data governance policies involverules around not only data usage, ownership,etc. but also around big data-specific datagovernance and data stewardship. MDM3cognizant 20-20 insightsFigure 3MDM and Social Architecture+TwitterCRMOthersPermissions/dataextractionlayerSocial conversationsare passed on to theanalytics tool.Entity information such asFacebook IDs, Twitter IDsand other extractable entityinformation is syndicatedto MDM.CompanyWeb siteFacebookPost-analysisattributes likenet-promoter flag,at-risk flag, Influencerflag etc., are passedon to MDM.Attensity, Pentaho, Google’s SocialInteraction Analytics, CognizantSocial Prism and many more.MDM-on-Cloud• Data cleansing• Data standardization• De-duplication• Workflows• Hierarchy, etc.Social Analytics Tool• Sentiment analysis• Lead generation• Customer acquisition• Net promoter identificationAnalyticsTools
  • 4. cognizant 20-20 insights 4needs to enforce policies and procedures thatensure:>> External data does not affect the integrity ofthe internal data.>> Data is relevant for the purpose for which itis mined.>> Escalation paths exist for data conflicts.>> Privacy policies are enforced.When Big Data and MDM Combine toAchieve a Business ResultAirlinesWith enormous customer acquisition costs,customer centricity is the key to profitability inthis industry, which makes airline carriers idealcandidates for big data programs.• Problem statement: An airline wants to sendtargeted promotions to brand advocates andinfluencers to increase coupon redemption andpositive word of mouth.• Solution:>> Route a social media campaign through theairline’s loyalty portal to capture customers’social media presence.>> Analyze big data to identify brand advocatesand influencers among these customers andtag them appropriately in the MDM hub.>> A downstream campaign management sys-tem can readily obtain this data from theMDM hub to identify brand advocates amongan organization’s customers and — based onthe brand advocates’ preferences, opt-ins,etc. — send targeted promotions to them.Multibrand RetailersDue to cost pressures, severe competition and per-ishability of certain products, couponing is amongthe most commonly used marketing tactics inretail. Targeted couponing enabled by big dataanalytics would help retailers improve couponredemption rates and increase profitability.• Problem statement: A retailer wants toidentify heavy users of certain products andsend them targeted coupons to enable quickinventory clearance.• Solution:>> Implement big data analysis software toanalyze point of sale data for loyalty cardholders, and tag frequent purchasers of aproduct in the MDM system as heavy users.>> A downstream couponing system can readilyobtain this data from the MDM hub to identifyheavy users of certain products and — basedon the preferences, opt-ins, etc. of these cus-tomers — send them targeted coupons.Fashion RetailersWith rapidly shifting styles, fashion retailers needto update their product assortment continuous-ly. Garments that are in vogue one season cango out of style the next, which means retailersmust clear their racks using discount pricingstrategies. The longer that old merchandise sitsin the warehouse, the more it loses value. So it’simperative for retailers to sell their merchandiseas quickly as possible.• Problem statement: A retailer wants tobuild a complete profile of all customers so itcan recommend ideal products and discountoffers, at the right time and place, to improvecustomer service and profitability.• Solution:>> Implement big data analysis software toanalyze point of sale data, purchase historyand data from social media (Polyvore, Pin-terest, Facebook, etc.) and create a completecustomer profile in the customer hub.>> Analyze data in the customer hub in realtime to recommend products, coupons, the time of checkout or as part of market-ing campaigns that send targeted couponsand promotions to these customers.Satellite TV and Cable ProvidersThese providers can analyze customer channel-surfing behavior for many purposes.• Problem statement: A satellite TV/cableprovider wants to show targeted ads tocustomers to increase advertiser ROI.• Solution:>> Analyze individual customer channel-surf-ing behavior to understand which customersare likely to watch which types of ads duringwhich shows.>> Enrich customer data in the MDM hub basedon this intelligence.>> Use downstream ad placement software toobtain channel-surfing information, alongwith viewer household information, opt-ins,etc. from the MDM hub on-demand, andbased on that insight, target ads to viewers.
  • 5. cognizant 20-20 insights 5Multiproduct Manufacturing CompaniesSuch companies must manage a large number ofsuppliers and have a 360-degree view of themvia a central repository. This is important notonly for supplier selection purposes but also forregulatory compliance.• Problem statement: A multiproduct manufac-turing company seeks a 360-degree view ofits global supplier base for quick and efficientsupplier selection.• Solution:>> Analyze available information about eachsupplier via the media, the Web and fromnotes transcribed by procurement manag-ers during transactions.>> Enrich supplier data based on this analysisto build a 360-degree view of each supplier,including violations cited in the news, delayin past shipments, etc.Automobile ManufacturersCar makers tend to spend millions of dollarseach year on market research for product lineextensions or feature enhancements. Big datacan be analyzed to understand what customers,partners and competitors think about a carand more.• Problem statement: In an attempt to expandits business, a car manufacturer seeks toaggregate all feature-related commentaryabout its models and use these insights toshape future vehicles.• Solution:>> Analyze social media commentary, calls tothe call center related to each model anddata from automobile monitoring systems,if available.>> Enrich the data about each car in the prod-uct information management system, in-cluding features to be added in the nextmodel, so that all information about the ve-hicle resides in the same place.Critical Move-Forward ConsiderationsWhile big data is already the next big thing in busi-ness-technology, and holds the key to achievingthe next level of customer centricity, organiza-tions that want to make sense of huge, prolifer-ating pools of structured, unstructured and semi-structured data need to keep the following inmind:• A big data strategy should be a companywideinitiative, with complete buy-in from seniorleaders.• Big data will impact not only MDM but almostall enterprise systems in any size company (i.e.,businessintelligencesystems,datawarehouses,ERP, CRM, campaign management, etc.).Therefore, a proper IT integration strategy iscrucial.• It is important to establish a clear understand-ing of the expectations from big data analytics.The social Web holds terabytes of data for mostbrands; acquiring, storing and attempting tomine all of this data is not only impossible butalso unnecessary.• Finally, brands need to maintain a balancebetween targeted promotions that consumersfind useful vs. targeted promotions thatconsumers may find intrusive or an encroach-ment on their privacy.Final ThoughtsIn this era of customer centricity, superiorcustomer service and operational efficiency aretwo major differentiators that can set a companyapart from its competitors. Big data analytics canhelp improve customer service, develop operation-al efficiencies and increase profitability. However,big data analytics cannot provide actionableinsights unless the data and related informationderived from big data sources can be traced backto the appropriate entity. Therefore, to get the fullvalue from big data initiatives, MDM is not just anoption but a critical necessity.About the AuthorReshmi Nath is a Business Consultant within Cognizant’s Customer Solutions Practice, focused onmaster data management. She has seven-plus years of experience in the technology and consultingindustries in the U.S. and India. Reshmi earned her M.B.A. in marketing and finance at Michigan StateUniversity and a bachelor’s degree in computer science at Bemidji State University. She can be reachedat
  • 6. About CognizantCognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process out-sourcing services, dedicated to helping the world’s leading companies build stronger businesses. Headquartered inTeaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep industryand business process expertise, and a global, collaborative workforce that embodies the future of work. With over 50delivery centers worldwide and approximately 162,700 employees as of March 31, 2013, Cognizant is a member of theNASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked among the top performingand fastest growing companies in the world. Visit us online at or follow us on Twitter: Cognizant.World Headquarters500 Frank W. Burr Blvd.Teaneck, NJ 07666 USAPhone: +1 201 801 0233Fax: +1 201 801 0243Toll Free: +1 888 937 3277Email: inquiry@cognizant.comEuropean Headquarters1 Kingdom StreetPaddington CentralLondon W2 6BDPhone: +44 (0) 20 7297 7600Fax: +44 (0) 20 7121 0102Email: infouk@cognizant.comIndia Operations Headquarters#5/535, Old Mahabalipuram RoadOkkiyam Pettai, ThoraipakkamChennai, 600 096 IndiaPhone: +91 (0) 44 4209 6000Fax: +91 (0) 44 4209 6060Email:­­© Copyright 2013, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by anymeans, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein issubject to change without notice. All other trademarks mentioned herein are the property of their respective owners.