turning big data into useful information

8/12/2019 Turning Big Data Into Useful Information

1/14a Storage eBook

Turning Big

Data Into UsefulInformation


2/14

2 The 4 Principles of a Successful Data Strategy

6 Transforming Information Into Knowledge

9 Building a Better Ball Team With Big Data:Lessons for Executives

11 Big Data, Big Opportunities for DBAs6

11

2

9

Contents

12

This content was originally published on the CIO Update, IT Business Edge and Enterprise AppsToday websites. Contributors: Paul Barth, Seth Earley, Loraine Larson and Susan Hall.

Turning Big Data IntoUseful Information


3/14

2 Turning Big Data Into Useful Information a QuinStreet Storage eBook. 2013 QuinStreet, IncBack to Contents


any Fortune 500 companies are recognizingenterprise data as a strategic business asset.Leading companies are using troves ofoperational data to optimize their processes,

create intelligent products and delight their customers. Inaddition, increased demands for regulatory transparency are

forcing companies to capture and maintain an audit trail ofthe information they use in their business decisions.

Despite this, large companies struggle to access, manageand leverage the information that they create in theirday-to-day processes. The rapid growth in the numberof IT systems has resulted in a complex and fragmentedlandscape, where potentially valuable data lays trapped infragmented inconsistent silos of applications, databasesand organizations.

ITs Not Your FaultExperience has shown that this is not a technologyproblem, it is a business problem. Creating an effectivedata environment requires change and coordination acrossthe board, with business and IT joined at the hip. To ensuresuccess, they must create a practical data strategy thatguides process changes as well as ongoing investments intheir data assets.

In our work with Fortune 100 companies during the

past 10 years, we have identied four principles behinda successful data strategy. These principles align andfocus the strategy, breaking initiatives into manageableprojects with a measurable business benet. Principles arepresented as questions that business and IT must answer.Those answers, in turn, help shape the framework andpriorities to drive implementation of the strategy:

Question 1: How does data generate business value? Improving the quality or accessibility of enterprise data

is not an end in and of itself. It is merely an enabler

for creating business value. The data strategy must bedriven by an understanding of how information canenable or improve a business process. For example,increasing cross-channel sales (a business value) requiresdata about your current customers and the productsthey own (the data); or reducing the cost of manualreconciliation for nancial reporting (the business value)requires standardizing and consolidating redundant andinconsistent data across business applications (the data).

The data strategy does not need to identify all possible

business benets, but it should dene several that arematerial to the business and measurable. Establishingsome early, visible benets is important to launching thedata strategy and giving it momentum.

Question 2: What are our critical data assets? Not alldata in the business is critical. In fact, most data is specicto an application, business function or transaction. Datathat is critical typically has two characteristics:

The 4 Principles of a Successful Data StrategyBy Paul Barth

M


4/14



It is associated with something of long-term valueto the rm (e.g., product, customer or nancialinformation)

It is used across multiple systems and businessprocesses.

A high-level process ow through marketing, sales,fulllment and nance for a top technology companywould go something like this:

Marketing creates interaction information as it reachesout to organizations and individuals through itsmarketing campaigns.

As sales leads emerge, a salesperson is assigned,partners are engaged and sales opportunityinformation is maintained throughout the sales process.

When an agreement is reached, terms are shared withproduct ful llment to deliver the product and maintainsupport.

Finance and sales validate commissions with the sales

teams and partners.

Management uses an end-to-end view of theseprocesses to evaluate the effectiveness of its pipelineand make ongoing improvements in and across theareas.

This process analysis reveals several critical data assetsand associated attributes. For example, customerorganization and individual information is used by everyone of the process steps. If this information is siloed and

inconsistent, customers will get inconsistent messages

and service. Process owners will have difculty measuringtheir effectiveness. Analyses will not reconcile. Andimplementing new controls or improvements will requirechanges within each process step.

Conversely, improvements to these critical data assetswill likely yield business benets in all ve areas.

ROI

In our experience, identifying and improving criticaldata assets in large companies can yield tens of millionsof dollars in benet, and justify millions of dollars ofinvestment in implementing a data strategy.

However, we believe it is just as important to keep theset of critical data assets as small as possible. Note thatthe most critical data asset for these subject areas is acommon identier. Maintaining the unique identity ofcustomers, products, interactions and contracts is whatlinks information across the enterprise. Once that istackled, attributes can be added incrementally to theenterprise record over time.

Question 3: What is our data ecosystem? For mostbusinesses, data is an active asset that is captured,created, enhanced and used in many business processesand applications. To manage this dynamic environment,the ows of data across systems and processes need tobe organized in a coherent way.

We use a business architecture (not a technologyarchitecture) to dene core data capabilities that businessand IT must create together. These capabilities organizetechnology platforms and business processes based on

their function in the ecosystem: capturing and creating

In our experience, identifying and improving criticaldata assets in large companies can yield tens ofmillions of dollars in benet.


5/14



data, cleansing and organizing it, mining business insightsfrom it, and using those insights to drive intelligentactions in the business.

By capturing data that measure the outcomes of ouractions, we create a closed loop that allows companies touse their data to test, learn and improve their processes.

The diagram below depicts three broad classes of corecapabilities: data, insight and action.

Data capabilities are responsible for creating andmanaging usable, high-quality enterprise informationassets. These

include all standard data management capabilities, suchas data sourcing and integration, quality and metadata

management, data modeling and data governance.

Insight capabilities include tools, data and processes formanagement reporting and advanced analytics.

Action capabilities provision data and businessintelligence to applications, business processes andbusiness partners, and capture responses to interactions.

This capabilities model can categorize thousands ofapplications and data repositories into 12 logical bucketsthat will guide their simplication and evolution toward acommon strategic blueprint.

Question 4: How do we govern data? Ultimately, theimplementation of a data strategy is not a project; it is anongoing function of the company that must be governed.Because data is so ubiquitous, the governance structuremust be federated, with a central governing bodyaddressing the most important, common data and mostof the data managed locally in the lines of business.

We have found several elements of this model criticalto successful governance.

First, the stewardship communityis business heavy, with

executive business dataowners supported

by business datastewards who reportto them. IT custodiansensure that the

systems incorporateand monitor therequirements of the

business.

Second, companies shouldincorporate data governance as a part

of other standard governance procedures asmuch as possible, including architectural review boards,audit and risk review processes, system developmentmethodology, and security processes. Over time, a

distinct governance body for data may disappear as it isfully embedded in other business governance activities.

Third, it is important to launch data governance with asmall facilitation team and some data governance relatedinfrastructure, such as data quality, metadata and lineagetools, to provide visibility and measures to the datagovernance board.


6/14



Conclusion

The alignment of the four principles for successful datastrategy is the foundation for establishing a manageable,meaningful change in the way companies deal with data.Note that technology is not the key to success it ismerely a supporting element in the development of corecapabilities.

For many rms, the rst attempt at a coherent datastrategy is a daunting effort, with stakeholders learningeach others language for the rst time. Over time,the common understanding of how data is vital to the

business establishes an effective dialogue so that trulystrategic initiatives can be launched that make everybusiness process more informed and intelligent.

The alignment ofthe four principles forsuccessful data strategyis the foundationfor establishing a

manageable, meaningfulchange in the waycompanies deal withdata.


7/14


8/14


9/14



of progress and success of the information managementinitiative.

The goal of good governance is to ensure the correctcapabilities are being developed and deployed, sufcientresources for success are allocated, policies are beingcomplied with and organizational change managementresources are engaged throughout the program.

These efforts require the goals of the program be fullysocialized, and users understand the end game andbenets as well as what is expected of them. Programssucceed or fail based on the level of socialization, buy-inand change management effectiveness.

5. Apply accepted practices to unstructured contentprocesses to promote better information hygiene Most enterprises have developed poor informationmanagement habits and practices. This is understandable,as the information tsunami is growing at an astoundingrate. IBM believes that in the next ve years enterpriseinformation will grow by 650 percent with 80 percent beingunstructured.

The only way to avoid the increasing inefciency andineffectiveness that this situation will create is to begin todevelop better information hygiene. This means:

Developing and applying information lifecycles.

Using the appropriate tools for collaboration versusreference.

Applying appropriate resources to organize importantinformation.

Assigning ownership and curation across repositories.

6. Measure the information management process Develop metrics to measure the quality and currency ofcontent. A big part of this is simply measuring success atgetting rid of materials cluttering up the system. This cansignicantly improve search results, make content moreusable, allow essential content to be prioritized, and ensure

that the curation process is focused on useful information.

Other useful metrics include how often critical contenthas been reviewed, whether tagging has been approved,the age of content, value ratings by users, heuristics foralignment with best practices, search precision and recalland, most importantly, statistical sampling of content forcompliance with governance process guidelines.

7. Measure business impact of new practices Thebusiness impact of content can be measured only throughits impact on a business process.

Content supports process (e.g., knowledge base contentsupports customer service reps) and the performance ofthat function can be linked to use of high-value content:Sales processes are supported by proposal content,competitive information, marketing content, and soon; and engineering is supported by various designlibraries, methodologies, test protocols and specicationdocuments, for example.

Content quality can impact a process which in turnsupports a business objective (e.g., customer retention,

customer acquisition and time to market).

By creating linkage to a specic step in a process andmeasuring how effective or efcient that step is withsupporting content versus the effectiveness with poorquality content, one can measure the impact of these newpractices.

8. Repeat on department by department basis Oncethis approach is initially developed and applied, create acenter of excellence with the purpose of introducing these

practices to other parts of the organization.

Content habits take time to develop. In fact, a baselinematurity assessment can tell you where the organizationis on the learning curve and how long it will take to movethrough the stages that are needed to achieve excellence.

While this journey will be challenging, there really is nochoice given the information avalanche that lies ahead.


10/14



more data been more persuasive? Hardly. Neither thequantity nor quality of data was the issue. What mattersis how and why vastly more data leads to vastlygreater value creation. Designing and determining thoselinks is the province of top management.

Specically, executive leaders need to decide whatmatters most and make a commitment to the desiredoutcome then decide how Big Data can help them getthere, he says.

This advice differs from what Ive heard previously, whichis that companies should approach Big Data as more ofan exploration to see what the data reveals, rather than aquery.

As a happy coincidence, European soccer shows both canbe true.

European soccer teams are super-aggressive about data.They dont monitor just the usual stats; they actually wire

Building a Better Ball Team With Big Data:Lessons for Executives

By Loraine Lawson

continued

xecutives are becoming a bitskeptical about Big Datas abilityto deliver real business change,it seems. The problem isnt the

skepticism the problem is that its based onmisunderstanding.

Dont believe me? Good. Ask a Europeansoccer coach or, if theyre in the UnitedStates, Oakland Athletics. Or, if you prefer,ask Michael Schrage, who rst noted theproblem during a Big Data seminar for seniorexecutives held in London.

Schrage is an author and research fellowat MITs Sloan Schools Center for DigitalBusiness. He asked the room full of executives,

How much more protable would yourbusiness be if you had, for free, access to 100 times moredata about your customers?

The response wasnt what he expected.

not a single executive in this IT-savvy crowd wouldhazard a guess, he writes in a recent Harvard BusinessReview blog post . One of the CEOs actually declared thatthe surge of new data might even lead to losses becausehis rms management and business processes couldnt

cost-effectively manage it.

Schrage contends this attitude is based on amisconception about becoming data-driven, but actually, itwill require more human decisions.

The reason why my London executives evinced littleenthusiasm for 100X more customer data was that theycouldnt envision or align it with a desirable businessoutcome, he writes. Would offering 1000X or 10,000X

E
http://www.itbusinessedge.com/cm/blogs/lawson/getting-real-about-big-data-the-people/?cs=50224http://www.itbusinessedge.com/cm/blogs/lawson/getting-real-about-big-data-the-people/?cs=50224http://www.itbusinessedge.com/cm/blogs/lawson/getting-real-about-big-data-the-people/?cs=50224http://blogs.hbr.org/schrage/2012/09/what-executives-dont-understan.htmlhttp://blogs.hbr.org/schrage/2012/09/what-executives-dont-understan.htmlhttp://blogs.hbr.org/schrage/2012/09/what-executives-dont-understan.htmlhttp://blogs.hbr.org/schrage/2012/09/what-executives-dont-understan.htmlhttp://www.itbusinessedge.com/cm/blogs/lawson/getting-real-about-big-data-the-people/?cs=50224http://www.itbusinessedge.com/cm/blogs/lawson/getting-real-about-big-data-the-people/?cs=50224


11/14



up players at practice to collect data in all biochemicaldata, GPS data and vital signs, reports InformationWeek.Coaches can actually create player-specic drinks andnutritional supplements for each player, based on bloodand saliva tests.

Heres the catch: Most of it is meaningless.

It turns out theres no correlation between how muchdistance a player covers on eld during the game andoutcome, the article notes. Likewise, traditional stats number of tackles, shots on goal or other popular gametrivia are useless. Those measures dont impact the

actual outcome of the game.

But by monitoring all this data, they did nd data thatcorrelated with winning, InformationWeek notes:

It took time for team managers and their statisticsgurus to understand which of those data points to payattention to. The percentage of completed passescompared to the number of interceptions is a goodindication of consistent victory, as is an error rate lowerthan 18 percent.

And, in the case of Oakland Athletics, they used the datato identify players who would be a unique asset to theteam, and exploit their opponents weaknesses, on asmall budget.

The moral of that story: The teams did not know whichdata they needed, but they knew the outcome theywanted to create.

The same will be true for businesses. Big Data can createbig outcomes, but youll have to identify the outcome youwant to identify the data that matters to your sport andyour particular teams.

Now, its true you shouldnt expect overnight wins.Theres actually a steep Big Data learning curve , as arecent CompTIA survey showed.

And thats not to say that a bit of skepticism isntdeserved there is a lot of hype now.

In a new era of Watson, Windows and Web 2.0

technologies, any organization that treats access to100X more customer data as more a burden than abreakthrough has something wrong with it, Schragewarns.

So dont dismiss Big Data as useless, just because itsbeing hyped. Instead, Schrage suggests CIOs andexecutives:

Rethink how your organization adds value.

Treat Big Data and the algorithms that run them likemanaging top talent.

Ask What value matters most, and what marriage ofdata and algorithms get us there? instead of how dowe get more value from more data?

The moral of that story: The teams did not knowwhich data they needed, but they knew the outcomethey wanted to create.
http://www.enterpriseappstoday.com/business-intelligence/steep-big-data-learning-curve-reveals-comptia-survey.htmlhttp://www.enterpriseappstoday.com/business-intelligence/steep-big-data-learning-curve-reveals-comptia-survey.html


12/14



recovery and all that other stuff. Thats part of theservice, he said.

Broader Denition of a DBA

Traditional DBA work still needs to be done, said NeilRaden, CEO and principal analyst at Hired BrainsResearch.

If you use a broader denition of a DBA as not so muchan administrator, but a data management person, thoseskills are going to be very valuable, Raden said, notingthat data management skills encompass a broad rangeof knowledge in areas such as data quality and dataintegration.

He said some of the most exciting job opportunities inBig Data involve statisticians and analysts who can work

Big Data, Big Opportunities for DBAsBy Susan Hall

continued

mid all the talk about Big Data, questionskeep popping up about how the databaseadministrator (DBA) ts in. The role ischanging and it varies by company,

according to Bert Scalzo, database domain expert forQuest Software.

One of the biggest concerns I keep hearing is howthe DBA role has either lost importance or gainedimportance. Theres been a change, and theresdisagreement about what the change is, he said.

DBAs More or Less Important?

Scalzo said some people feel the DBA role is becomingmore important, due to the large amounts of datacompanies are accumulating and storing, thanks to

inexpensive disk space and cloud storage. That doesntmean there are more DBAs, however.

It used to be that you could only have so manyterabytes or petabytes before it went to secondarystorage. Now thats kind of old-fashioned. Now youcan keep data live until theres absolutely no reason tohave it, he said. In that case, the DBA is doing more.Its not that the number of DBAs is decreasing, but theamount of data that companies are keeping is increasingexponentially with the same amount of resources.

Those people see DBAs growing more important, but atthe same time more overwhelmed, he said.

Some other companies treat data as a commodity andmay think they dont even need a DBA on staff. Suchcompanies are more likely to outsource DBA work.

If its in the cloud, Amazons making sure it performswell, doing the tuning, optimization, backup and

A


13/14


14/14

turning big data into useful information

Documents