To Find the Value in Your Data, Stop Searching for It

To Find the Value in Your Data, Stop Searching for It.pdf
of 4
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Podcasts Use Cases University Map Glossary Whitepapers Webinars Guides & eBooks ManagementSkillsVisualizationsPredictive AnalyticsMarketingOperationsCloud Christer Johnson, Principal, Ernst & Young RReellaatteedd SSttoorriieess IBM: Firms Looking to External DataSources to Elevate Analytics Value. Read the story »  ( Domain Expertise Vital to Unlocking To Find the Value in Your Data, StopSearching for It by Christer Johnson (  | April 11, 2014 5:30 am | 1 CommentsMobile devices, smart phones, and devices used in customertransactions deliver some 5 billion gigabytes of information tocorporate databases every day. Naturally, many corporateanalysts and sales and marketing teams are inspired by theopportunities to mine the data to gain valuable insights aboutcustomer behaviors and preferences.Starting with a large pool of data, paradoxically, can actuallymake it harder to find value. To find it, companies shouldn’tfocus on how to derive value from data, but rather on thebusiness decisions that the company can make moree ! ciently and e ectively. To use an explorer analogy, there isa reason explorers who set out with a clear goal ordestination in mind tend to be the ones we remember.Columbus didn’t set out to “get value from the ocean.” Hesought an ocean route to Asia and ran ashore in the Americas along the way, ultimately gaining awareness of alarge land mass between Europe and Asia to the west.Once a key business challenge has been identified and itcomes time to generate value from big data, companiesshould proceed sequentially through the following three steps in the analytics value chain: Collect and manage data e ectively1.Perform sound and appropriate analytics2.Use the results from steps 1 and 2 to drive the key business decisions3. Before actually collecting and analyzing any data, however, it is important to set the stage forobtaining value by thinking carefully about what decisions you need to drive in step 3, and then workbackward to map out the type of analytics and data that would support those decisions. Experiencehas shown us that the companies that derive the most value from data and analytics routinely start byasking, “What decisions need to be made better and faster?” In other words, they start with the endin mind and then figure out which analytics need to be performed and which data might be relevant.Obtaining value from analytics is about having the right insights, at the right time, and in the rightplace to make impactful decisions. It is not just about producing interesting information.Let’s look at how one telecommunications company was able toshift its data collection and analysis approach to support thisstrategy. An analyst at the company developed a churn modelwith marketing data to predict the likelihood that customerswould drop their accounts within 15 days. The model generatedan accuracy rate of almost 85 percent, but the 15-day warningo ered the company insu ! cient time to prevent customers fromleaving. As a result, the highly accurate model did nothing to JJooiinn tthhee CCoonnvveerrssaattiioonn  ( /cs/c/?&cta_guid=74807e22-84ef-4e77-baa9-1551b00e72d6& placement_guid=659de830-867b-439a-809e-872d132cc6ec&portal_id=215445& redirect_url=El9oj%2B0vrUd6hNDNVQ%2B /yyFEuHEB6CO1bXw7W /8%2BY%2Bi0CnljQ3oqiHFaPVjSCplouKat%2Biv=ydENuXx3pWw%3D&hsutk=& ( /cs/c/?&cta_guid=e912b2ca-34 -4226-829c-40a28d7bea07&placement_guid=ae05c418-e4a9-484a-898f-7ec0af3fdc05& portal_id=215445& redirect_url=poUj59lKV17Zko4TFxbDR7k8RIjcdu /547hd5jMQJCzidCJ /x97pv2NonImAy44Sh5zG0h /%2BFsA6UkDnqPybSZcd3TMtVWQo8BYiVK /3o0QBptm0Jp8viYUpHbE%3D& iv=9YEr6y8x5BE%3D&hsutk=& (htt   p:// /cs/c/?&cta_guid=cbc6249e-9d39-4ab8-951b-35cfaa5a04e6& placement_guid=8f45ef68-49df-4455-8a88-5222115317da& portal_id=215445& redirect_url=MzJwdJJPXjl6aqfUA7qMvzmOGNDfiv=i/X3WQhuqIc%3D&hsutk=& ( /cs/c/?&cta_guid=bdbf54b8-57e2-473a-8104-6a5a3656fb3d& placement_guid=738a0fcd-e7aa-498b-9d96-98f31426118c&portal_id=215445& redirect_url=CS4jwCy8%2Bo2kfA  To Find the Value in Your Data, Stop Searching for It of 46/4/14, 11:36 PM  Value of Machine and Mobile Data. Read the story »  ( Government Data Chiefs Find VariedMeasures for Analytics Value. Read the story »  ( The Challenge of Aligning HRSurveys and Data with BusinessValue. Read the story »  ( help the company achieve its desired goal of reducing attritionbecause the company didn’t have time to make businessdecisions that could impact a customer’s choice to leave or stay.Several months later, a new analyst attacked the attritionproblem from a di erent perspective. He realized the goal was topredict, as soon as possible, customers at risk of leaving. Thecompany required time to make decisions and take actions tokeep the customer – even if it meant predicting churn with alower overall accuracy rate. By starting with the key requirement,that the company needed more advanced notice regardingcustomers likely to churn, the analyst realized he should explorenew and di erent analytical techniques and data sources.In relatively short order, the analyst decided to talk with theoperations unit of the company to see if data existed on droppedcalls and when the coverage dropped from 4G to 3G. The datawas available, and he quickly gained access to it and was able to merge it with the existing customerdata used to create the srcinal model. Using the new data, the analyst built a model that identified,with only slightly lower accuracy than the first model, customers likely to leave within 60 days. Thatadditional lead time gave the company more time to take action and design programs and o ers thatsignificantly improved customer retention. In other words, the company clarified the necessary timingof the decisions it needed to make and then built an analytical model to meet that timing.One way to prioritize the business decisions that deserve a more rigorous analytic approach is toadopt a more exploratory data-mining approach centered on identifying what drives a specificbusiness metric, such as revenue or cost. This approach di ers from the traditional method of forminga (narrow) hypothesis and then looking to either prove or disprove that hypothesis. The traditionalapproach increases the chance of missing critical insights into what truly drives revenue or costbecause the potential drivers are too narrowly defined. Similar to the di erence between a multiple-choice question and a free-form answer, the former gives an answer but the latter allows for newthinking.One company, which recently analyzed its sales and pipeline data, had long held a hypothesis thatrevenue was directly related to pipeline. As it turned out, there was minimal correlation betweenpipeline and revenue because the company had no way to measure or represent the quality of eachsales opportunity entered into the pipeline. As the company put more pressure on the sales force toexpand the pipeline, sales resources began entering lower quality opportunities to meet the demand.Using exploratory data-mining techniques to identify which deals had a higher win probability, thecompany determined that the win rate when the “competitor” field in a pipeline entry was left blank orunknown was close to 5 percent. That is much lower than the 30 percent win rate on entries in whichthe competitor field was filled in. Thus, exploratory data mining helped the company rethink along-held belief that encouraging sales people to expand pipeline is a key driver of sales. Subsequentanalysis was designed to support a more nuanced approach to drive pipeline quality and to prioritizee orts on the deals with the highest probability of winning.Once the company realized that pipeline was not strongly correlated with revenue, management’seyes were opened to explore other factors that drive changes in revenue. One of the factors theyidentified was price discounts o ered at the end of the quarter. The company realized that 30 percentof its sales were tied to a product for which the price was not significantly correlated with changes insales. The company’s previous emphasis on providing quarter-end discounts had trained itscustomers to wait until the last minute to receive the discounts the company o ered. This analysishelped the company make better decisions about when to lower price and when not to lower price.Companies that can consistently find the most e ective ways to extract value from their data will bethe ones positioning themselves to win in the global marketplace. Companies that don’t approachdata with an end goal of finding insights that help them solve specific business issues likely will findthemselves floating in an ocean of big data, wasting resources and, even worse, losing theircompetitive edge. Christer Johnson is a Principal within Ernst & Young’s National Advisory Practice  ( /US/en/Services/Advisory/Risk)  , Enterprise Intelligence, Analytics. He brings 20 years of experience leveraging advanced analytics and mathematical techniques to help his clients improve the e !  ciency  and e   ectiveness of business decisions and processes across industries including government,transportation, consumer package goods, healthcare, insurance, life sciences, and telecommunications.  /Cua0kscaHvHl1c0sieHCkCqXEh4Aaszazl0jS36Y%3D%3D&iv=km9v6rL1NeY%3D&hsutk=& ( /cs/c/?&cta_guid=12cf3789-4e3a-47ba-b05b-b08b9d8e89c2&placement_guid=c23e3344-f4fa-4f06-9822-b8933756f120& portal_id=215445& redirect_url=fHqYmiIhBEekQfwAgwetmFia0mm3iv=uCoqHl%2BsJBE%3D&hsutk=& SSeeaarrcchh   MMoosstt PPooppuullaarr AArrttiicclleess Use Predictive Analytics to Improve the CustomerExperienceTake a Data-Centric Approach to Managing YourUnstructured DataCincinnati Bengals Take a ‘Moneyball’ Approachto Talent EvaluationData Informed’s Top 10 Stories of 2013Can Big Data Grill the Perfect Burger? To Find the Value in Your Data, Stop Searching for It of 46/4/14, 11:36 PM  ( 77-4565-867f-1a8c909bf112) Tags: analytics ( , big data ( , data management ( , data science( One Comment  Sam AustinPosted April 17, 2014 at 6:10 pm | Permalink Bravo, this is spot on. Without questions to guide inquiry, and data that is collected, cleansedand segmented with a documented process, scrupulously mindful of its potential uses andlimitations, data can’t tell us anything that would justify the time and e ort it takes to analyze it,and we can set little store by what it does tell us. Yet too many still seem to think that dataanalytics can answer all their questions without the necessity of formulating even one, and thatsome magic button must exist that will both reverse the ills of prior neglect and obviate thenecessity of rigorous and consistent data hygiene. Stop the madness! Reply1. Post a Comment Your email is  never   published nor shared. Required fields are marked * Comment Name*Email*WebsiteNotify me of followup comments via e-mail Most Recent Posts Share 40   Data Informed + 680 Follow +1 Meet the Staff and ContributorsMeet the Staff and Contributors   To Find the Value in Your Data, Stop Searching for It of 46/4/14, 11:36 PM   About UsContact UsAdvertisingNews RoomPrivacy PolicySite mapTerms and Conditions 20 Carematrix Drive, Dedham, MA 02026, USA | Sales and Customer Service: 1.781.751.8799 | Email: ( © 2012-2014 Wellesley Information Services. All rights reserved. MapR Looks to Enhance Hadoop Accessibility with App Gallery ( Correlating Sales Data with Customer Behavior Data to Improve Sales and CustomerInteraction ( An Exec’s Advice to Data Analysts for Getting Heard in the Boardroom ( /executives-advice-analysts-getting-heard-boardroom/)Cincinnati Bengals Take a ‘Moneyball’ Approach to Talent Evaluation ( /cincinnati-bengals-take-moneyball-approach-talent-evaluation/) To Find the Value in Your Data, Stop Searching for It of 46/4/14, 11:36 PM
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks