BIG DATA IS HERE to stay. But like so manybusiness trends, the term is often misused,misunderstood and misapplied. Understandingexactly what big data is (and isn’t), theadvanced analytic solutions available toleverage it and the imagination to apply dataand analytics to address critical confectionindustry questions open new and excitingopportunities to achieve significantimprovements in revenue, profitability andinventory management.

Successful applications of big data havea few things in common:• Securing highly granular data; e.g., store-level data or individual shopper data.

• Collecting different types of interrelateddata — the most interesting applicationsof big data come from the integrationand correlation of different types ofhighly granular information, e.g.,household demographics at the ZIP-codelevel and store-level sales intel.

• Applying analytic techniques that workwith incomplete and imperfect data sets;data collected from multiple sourcesoften have holes and might include badsets.

• Thinking creatively; big data can addressissues that previous analytic solutionscould not. Marketers should thinkbroadly about interesting applicationsthat can drive value.

SORTING THROUGH THE HYPETechnology analyst firm Gartner Inc. offers aclear definition of big data: “High-volumevelocity and variety of data/information assets,

structured and unstructured, that . . . can beanalyzed in real-time with innovative and cost-effective techniques and technologies . . . forenhanced insight and decision making.”

To put the amount of data available todayinto perspective, Americans produce 2.5 milliongigabytes of data each day, which is projectedto grow 40 percent every year, and 90 percentof the world’s data has been generated in thepast two years.

Today, 84 percent of manufacturers areaware of big data as a concept, but just 36percent know the amount of unstructuredinformation within their organization. To date,16 percent of manufacturers have executed abig data project, 24 percent are currentlyimplementing one and an additional 28 percenthave a strategy in place.

DEMONSTRATING BIG DATA’S POWER To illustrate the business opportunities big dataaddresses that can lead to enhanced growth,following are three scenarios confectionerymanufacturers and retailers often face, alongwith a discussion of the data and analytictechniques that lead to better decisions anddrive growth.

Big data isallowingcompanies acrossthe supply chain to use a laser focuson the mostopportunisticareas for growth. InformationResources, Inc.Chief InformationOfficer Ash Patel,discusses how touse it to effectivelyreach yourcustomers.

Getting Big DataTo Work For You

26 Candy&SnackTODAY J a n u a r y / F e b r u a r y 2 014 w w w . c a n d y a n d s n a ck t o d a y. c o m

MARKET INTEL

CONTINUED ON PAGE 28

Onlyof manufacturers

have executed a big data project.

16%

CST_14-1 p026-030 BigData_Layout 1 3/7/14 9:02 AM Page 26

SPECIAL REPORT

CONTINUED FROM PAGE 26

‘Americans produce2.5 million gigabytesof data each day,which is projected togrow 40 percent everyyear, and 90 percentof the world’s datahas been generated in the past two years.’ASH PATELInformation Resources, Inc.

28 Candy&SnackTODAY J a n u a r y / F e b r u a r y 2 014 w w w . c a n d y a n d s n a ck t o d a y. c o m

MARKET INTEL

OPPORTUNITY #1: CONSUMER ACTIVATIONBig data, applied effectively, gives marketers the opportunity to target consumers with highprecision through “one-to-one” marketing and promotions focused on products they careabout at the retail outlets they shop through the entire path to purchase.

Data sets should come from a variety of sources, including national consumer panels,independent database sources, such as Experian and Acxiom, U.S. census information, socialnetworks and related data.

Integrating these structured and unstructured data establishes a holistic picture of theshopper. Applying consumer propensity scoring models will assess the likelihood of consumersto buy a product or brand and measure their affinity for specific media types. It will also enablesegmentation and targeting based on online and offline browsing and shopping behaviors.

These analytics enable marketers to target consumers individually or in segments with theappropriate products through suitable media. Dashboards driven by these analytics results willenable marketers to monitor sales performance relative to penetration targets and media ROI.

The impact on consumer activation of an effective big data implementation is profitgrowth in the five to 10 percent range in segments and categories where individual shoppersare targeted.

CASE STUDY: HYPER-LOCAL TARGETING, DIGITAL MARKETING ACHIEVE SALES INCREASES

A powdered drink mix manufacturer was able to generate a 5.5 percent dollar salesincrease and expand household penetration by 11 percent, achieving a $2.65 return for $1of advertising spend.

The company’s marketing team achieved this by collecting online data such aswebsites visited, media consumed, coupons downloaded and social media activity. Theyalso integrated consumer panel data, including purchasing behavior, attitudes, shoppingbehavior and demographic information. Finally, they included POS data by store.

With this information, marketers analyzed category development (CDI) and branddevelopment (BCI) indices of the company’s product and competition using panel andPOS data. This identified consumers and households where the drink mix wasunderdeveloped relative to the category. The team then constructed look-alike modelingto map and identify target consumers on the digital partner’s database.

As a result of the analysis, the marketing team was able to recommend consumersand households to target for a new digital marketing campaign. This recommendationwas based on actual sales and consumer buying behaviors and mapped to the digitalcampaign partner’s database.

GETTING A CLEARER PICTURE OF CONSUMERSThe ability to capture more data on consumers from disparate sources

can significantly improve manufacturers’ and retailers’ understanding of target shoppers’ attitudes and behaviors.

• URBAN DWELLER

• 18-35 YEARS OLD

• MID- TO UPPER-INCOME

• ETHNIC MINORITY

• FINANCIAL SERVICESPROFESSIONAL

• RETURNS 40% OF WHATSHE BUYS ONLINE

• EXCITED FOR HAPPYHOUR TOMORROW!

• CURRENTLY INMANHATTAN NEAR 55TH AND 7TH AVE.

• UPSET WITH RECENTEXPERIENCE AT UNITEDAIRLINES; LET ‘EM HAVEIT ON FACEBOOK ANDTWITTER

small data BIG DATA

CST_14-1 p026-030 BigData_Layout 1 3/7/14 9:02 AM Page 28

CONTINUED ON PAGE 30

‘New analyticsalgorithms enablemarketers to bothpredict future market trends and prescribe specific activities to take.’ ASH PATELInformation Resources, Inc.

MARKET INTEL

w w w . c a n d y a n d s n a ck t o d a y. c o m J a n u a r y / F e b r u a r y 2 014 Candy&SnackTODAY 29

Big data develops the opportunity for marketers to accelerate newproduct launch timetables and success rates through theintegration of social, mobile and online data feeds with moreclassic consumer data and research techniques. It also enablesbuilding strong brand commitment through social and onlinecommunities to drive new business success.

Data required to address this opportunity are collected fromsocial media and data sites, including websites designed forspecific consumer needs, such as babycenter.com, and relevantblogs, interest sites, intent sites and apps, POS data, consumerdata collected from panels or databases, such as Experian andAcxiom, and brand tracking research.

Next, marketers analyze the social, online and mobile dataand integrate it with classic consumer research. They will consider

crowd-sourcing product ideas to gain additional consumerfeedback, analyze and track new products over time, forecast newproduct launch performance and implement productrationalization and portfolio optimization analytics.

In addition, marketers will monitor and track consumersentiment through metrics such as “likes” or tweets, and leveragesocial, online and mobile communities of users to build deepconnections and commitment to the brand. They will gain newproduct launch ideas and performance tracking.

Big data support of product innovation can result in a five to10 percent improvement in new product launch success rates,speed of ideation, course correction and monitoring ofcompetitive products.

CASE STUDY: DRIVING IN-HOME COFFEE INDULGENCEA leading coffee system manufacturer and a coffee productmanufacturer came together to make an at-home coffeeindulgence offering that rivals café-style servings. Successfulleveraging of big data resulted in growing the category 24percent in its first two years through innovation, and itcontinues to be among the fastest-growing CPG categories.

The marketing team began by qualitatively andquantitatively evaluating the size of the market opportunity.They secured hardware sales data to understand consumeracceptance of the equipment for brewing the coffee products.They secured POS data to support base sales and promotionand obtained panel data to understand adoption, repeat salesand incrementality.

The analytics applied to these data included focus groups;concept and in-home product testing; tracking adoption ofboth hardware and coffee product; trial and repeat builds;source of volume analysis; interviews with consumers tryingthe product to understand opportunities and barriers toacceptance; and finally, category measurement to assesscannibalization or incrementality.

As a result, marketers were able to execute enhancedtargeting of early adopters via look-alike modeling, make dailyversus special-occasion product mix recommendations andoffer channel coverage and retail execution recommendationsacross brick and mortar and e-commerce channels.

OPPORTUNITY #2: PRODUCT INNOVATION

POTENTIAL BIG DATA APPLICATIONS FOR CPG COMPANIES

CATEGORY BUSINESS ISSUESSALES

GROWTHPROFITABILITY

BRANDHEALTH

SALES &MARKETING &EFFICIENCY

MARKETRESEARCH &EFFICIENCY

DEMAND -SUPPLY

OPTIMIZATION

VALUE DRIVERS

- Shopper profiling and segmentation- Shopper marketing- 1:1 offers- Sequencing promotional activity

- Idea generation- Product and packaging testing- Usage tracking- Product feedback

- Granular (1:1) spend allocation- Real time re-allocation of funds basedon market feedback

- Measuring ROI of social campaigns

- Demographic & geographic assortment- Co-branding/co-marketing- Online/sentimentality effects

- Cross-channel price optimization- Price architecture aligned toretail/consumer value

- Dynamic pricing strategy

- Distribution voids- Out-of-stocks- Product freshness- Optimize inventories and productionschedules

Consumer &Shopper

Activation

ProductInnovation

MarketingMix

AssortmentPlanning

PriceManagement

RetailExecution

X X X

X X X X X

X X X X

X X X X

X X X X X

X X X X

CST_14-1 p026-030 BigData_Layout 1 3/7/14 9:02 AM Page 29

SPECIAL REPORT

CONTINUED FROM PAGE 29

‘The impact onconsumer activationof an effective bigdata implementationis profit growth in the five to 10 percent range.’ASH PATELInformation Resources, Inc.

30 Candy&SnackTODAY J a n u a r y / F e b r u a r y 2 014 w w w . c a n d y a n d s n a ck t o d a y. c o m

MARKET INTEL

OPPORTUNITY #3: RETAIL EXECUTION

Managing distribution to minimize out-of-stocks and maximizeproduct freshness and quality is an area where effective big datacan make important contributions. Big data will also enhanceplanning for promotional events and maximize season- andevent-driven opportunities. Marketers should collect daily POSdata down to the store level, as well as inventory, shipment,weather, sales promotion and other causal data.

To study the data use modeling of expected sales by store,brand and product, as well as seasonal changes and expectedimpact of promotions. Apply algorithms to predict out-of-stocksby comparing daily actual sales to expected sales. Completing

audits and adjustments will improve these models.This analysis will result in decision support solutions, including

alerts and exception reporting on products and stores with out-of-stocks, quantification of the market opportunity, prioritizationand organization of issues based on the size of the opportunity,and weekly reports identifying the top 10 issues to be addressed.

Effectively applying big data to inventory management canresult in a sales increase of one to 10 percent, depending on thescale of the market issue or opportunity, as well as fulfill thepotential of seasonal, sports and other events.

CASE STUDY: STORE-BY-STORE ASSORTMENT OPTIMIZATIONBrand managers at a multi-product, global manufacturer wereeager to optimize their assortment to maximize sales potential,but lacked the data to make optimal store-level decisions. Themanufacturer offered more than 1,000 products, but each storewas only able to stock approximately 100 items and brandmanagers had to address more than 200,000 stores in total.

They began by collecting consumer panel data, seeking outbrand preferences and purchase behavior in the company’scategory. They integrated POS store-level data forapproximately 15 percent of the stores and included companyfinancials, shipments and planograms, as well as other store-level data, such as shopper profiles and store attributes.

To analyze the data, stores were clustered based onconsumer profiles and sales. Analyzing and integratingavailable, patchy data and making a complete store view

imputed missing data. Distributor limitations, account policies,space and brand strategies were factored into an assortmentmodel. Finally, a market structure simulation that providedincrementality and demand transfer implications of assortmentdecisions was developed.

This enabled brand managers to optimize assortment bystore, conduct scenario planning and modeling to test differenttrade-offs, and fill gaps in data using advanced imputation andextrapolation algorithms. They also gained the ability to makeretailer profiles to identify trends and attributes that separatestores, work with wholesalers on route planning to ensuremaintenance of assortment, and avoid out-of-stocks.

As a result, the company achieved a six percent sales liftand a $7 million per month increase in profitability across storeswhere the big data application was implemented.

Big data is in its third evolution. In the firstwave, decision makers were able to secure largeamounts of data, but lacked the powerfulanalytical tools to wring out critical insights.

In the second phase, large amounts of datacombined with improved analytical solutionsenabled marketers to gain value insights, butonly focused on past activities.

In the current phase, new analyticsalgorithms enable marketers to both predictfuture market trends and prescribe specificactivities around products, promotions,assortments and other key questions — anespecially exciting new development.

STRATEGIES FOR MOVING FORWARDWhile each big data application should reflectthe specific issue to address, common steps are:

•Brainstorming potential big dataapplications to address the company’sstrategic priorities.•Securing the data sources necessary toconduct the project. These can and shouldinclude internal and external data sources.•Identifying the analytic techniques required. •Developing the application and testing itsfunctionality, quality of results and ROI.

•Deploying the application to address thebusiness issues at hand.While this process might sound daunting —

especially if the application marketers would liketo develop is complex — harnessing big data canreally be much simpler than it sounds.

As the volume of data continues to growand analytic solutions improve, the ability toeffectively segment, target and activateconsumers is limited only by the imagination ofmarketers. Those who apply big data mosteffectively will enjoy sustained improvements inrevenue, profitability and inventorymanagement. CST

CONTRIBUTOR’S INFO

Ash Patel, chiefinformation officer forInformation Resources, Inc.,has more than 20 years ofexperience in theinformation technologyfield and currently isfocused on generating

value from technology solutions. He can bereached at ash.patel@iriworldwide.com

CST_14-1 p026-030 BigData_Layout 1 3/7/14 9:02 AM Page 30