Big Data Strategy | 285
social media sentiment of its customers, and some businesses are well-established with
stable clientele and low competitive worry. Hence, these kind of insights that BigData
can provide around how to acquire new customers or how to retain old ones may not
provide that much value to those businesses. In reality, most businesses these days
probably do not fit in these categories, but there are still a number that do. For those
that do, it really does not make sense to embark on a Big Data journey, if the benefit
that is going to come out of implementing a Big Data technology is of relatively small
value to the business.
e. Is ‘Big Data’ really essential for decision making?
Even if one assumes that one does need Big Data, let us consider the fact that ‘big’ is a
relative term. Business Intelligence or Analytics will provide value irrespective of whether
it is run on so called Big Data sets, or whether it is run on more modest or transactional
data sets. Established technologies like relational databases and BI technologies that have
been around for a couple of decades and more, can actually handle very large volumes of
data. Again, if one thinks that one does not need Big Data, then it is important to consider
the fact that even small businesses may have click stream or sensor-based data or other
BigData that they can take advantage of.
So, it is important to ensure that if one decides that one does need Big Data, the data
one is dealing with is Big Data in the industry sense of the term, encompassing a number
of criteria and not simply an amount of data that seems big.
11.5 GETTING READY FOR A BIG DATA PROGRAM
Big Data programs offer an enterprise a number of opportunities as we described and it can be
summarized into the following as depicted in Figure 11.4.
However, even though when Big Data promises to address some of the most important ques-
tions before the modern enterprise, executing a Big Data program is much harder than what a
number of organizations predict.
The illustration as depicted in Figure 11.5 summarizes most of the reasons why Big Data pro-
grams fail. In the subsequent sections, we shall address these issues.
• Choice of the right technology platform, keeping the aspects of commercials and human
resources in mind.
• Choice of the right Business Intelligence and/or Analytics platforms.
• The right infrastructural decisions around whether one needs to opt for Cloud-based plat-
form or on-premises solutions.
• Whether it is enough to extend or re-architect existing capabilities, instead of embracing
whole new technologies.
However, very often, just below these problems on the surface lies the single-most important
reason for the failure of Big Data programs, which is nothing but poor quality of data.
M11 Big Data Simplified XXXX 01.indd 285 5/13/2019 9:57:45 PM
286 | Big Data Simplied
FIGURE 11.4 Opportunities from embracing Big Data
Big Data opportunities
Products sold based on real time demands
Understand customers and relationships with them
Make better product recommendations
Understand multi-channel customer interactions
Real time analytics for risk management
Optimize inventory management
Manage suppliers better
FIGURE 11.5 Reasons why Big Data programs fail
Reasons why Big Data
programs fail
Poor planning resulting in delays
Unclear program objectives
Lack of technical talent in data
Lack of technical talent in analytics/business
intelligence
Incorrect platform choice
Exceeding projected costs
M11 Big Data Simplified XXXX 01.indd 286 5/13/2019 9:57:46 PM
Big Data Strategy | 287
The illustration as depicted in Figure 11.6 summarizes the symptoms of bad data and they are
as follows.
a. Duplicated as data for a particular business entity, like customer, product, supplier, employee
comes from multiple sources.
b. Inconsistent as the data is stored in different formats.
c. Incomplete as data entry is not verified in applications.
d. Erroneous as the data was never cleaned to ensure accuracy.
e. Unreliable for all of the above reasons.
A number of organizations fail to realize how unprepared they are from the perspective of
data quality before they jump into a Big Data or Analytics program. Therefore, regardless
of the sophistication of technology type and the accuracy of decisions around Big Data or
Business Intelligence platforms, the programme is bound to fail. It gets worse when an
organization tries to overcome this problem through ad hoc, often manual solutions like
manual xing of data errors. Eventhough when such methods are sufce for a smaller scale
or even for a pilot implementation, it will never work for a large-scale Big Data programme
implementation.
The illustration as depicted in Figure 11.7 summarizes the impact of poor data on Big Data
programs.
Therefore, one of the prerequisites of a Big Data programme is to use tools for data cleansing,
such as Informatica, IBM, Talend, among others.
FIGURE 11.6 Symptoms of bad data
Symptoms of bad data
Duplicated
Inconsistent
Incomplete
Erroneous
Unreliable
M11 Big Data Simplified XXXX 01.indd 287 5/13/2019 9:57:46 PM
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.