Analytics Part 2: Beyond the Zetabyte


Anyone not living under a rock in the eDiscovery space has heard that Enterprise data is doubling every 18 months. The computing processor power is doubling every 18 months. The memory and storage cost is halving every 18 months (See Moore’s Law). The increased effort to derive insights out of this scenario is called the interpretation of data. And that is where Analytics step in and play a crucial role.  This is not just true in the eDiscovery space, of course, but it permeates across nearly every business vertical.

Take the example of something as seemingly common place as retail or food services.  A wide-array of analytics can apply to:

  • Developing close relationships with customers based on a deep understanding of their behaviors and needs;
  • Delivering targeted advertising, promotions and product offers to customers that will motivate them to buy;
  • Balancing inventory with demand so you’re never out of stock or carrying excess inventory;
  • Charging exactly the price that customers are willing to pay at any moment;
  • Determine the best use of marketing investments;
  • Locating stores, distribution centers, and other facilities in optimal locations.

The application of analytics is not merely a matter of business efficiency, it is one of major cost reduction and revenue increase. For example, CVS uses analytics to target coupons at the point of sale and view its analytical capability as a nine-figure profit center. Hudson’s Bay corp. of Canada broke up a $26 million fraud ring with one analytical application. UK-based retailer Marks & Spencer achieved $2.5 million in labor efficiencies and $1.5 million in operating improvements (More Here), as well as a number of intangible improvements, through planogram automation and optimization, (Analytics driven shelf optimization).


The application of analytics to customer data with respect to segment, target, and personalize offerings is commonplace among entities that are household names.  The ubiquitous “Club Card” that companies from Ikea, to Gold’s Gym, Safeway, Costco, Best Buy, Walmart and Target lure customer’s to use with “Huge Savings” are massive sources of big valuable data that statisticians, analysts and mathematicians work feverishly to analyze and use to maximize profits.

big data

These programs are hugely successful. For example, Nieman Marcus believes, for example, that the top hundred thousand customers in its complex loyalty program, InCircle, account for almost half of its total revenues. Top customers can win free fur coats and even a Lexus. uses targeted and event-driven email marketing to refine category and product offers and even pricing. Targeted email offers has led to a 50% revenue lift from the email channel, and average order size increased by 6%.

Retailers with a web presence (Amazon, CV, HBC) use analytics to predict situations where Fraud is likely to occur and prevent it. has an aggressive program to predict and prevent credit card fraud, led to 50% reductions in fraud after just 6 months.  The system at has reduced fraud by $26 million since installation.

Walmart, uses analytics to determine everything from pricing, to inventory, store layout, security and design. Stocking patterns are based on actual consumer purchases, area demographics, preferences from consumer surveys, and inputs by local store managers. Individual managers can specify important local events and the beginning and end of local seasons, e.g., for hunting. the company’s Retail link portal is used to pass on local store assortments and replenishment needs to manufacturers. (Read more here Realizing the Potential of Retail Analytics)


When looking at the analytics industry one sees the application of of these tools to various industries. This includes web analytics, search engine optimization, artificial intelligence/data mining, and social network analysis.

Web Analytics

The most commonly known form of analytics is web analytics, thanks to the widespread use of tools such as Google Analytics that analyze and report on web page visits.  Web Analytics is being increasingly used alongside Social Network Analysis and Data Mining for Customer Analytics to provide integrated cross-channel market intelligence, campaign management.

Search Engine Optimization (SEO)

SEO is focused on and informed by the questions that Web analytics answers:

  • Which pages do people visit?
  • How does this change with date and time?
  • Where do visitors come from (geographical)?
  • Which site linked to us?
  • What were the search terms that led people in?
  • Is my site user-friendly?
Artificial Intelligence /Data Mining

Artificial Intelligence (AI) has something of a sci-fi reputation but AI research has produced some practical methods to allow machines (computers) to learn, in a limited sense of the word. Some of these methods became the basis for data mining, also known as “knowledge discovery in databases”. Through these so-called “machine learning algorithms”, computers are able to detect patterns in streams of complex and potentially incomplete input data. 1

Social Network Analysis (SNA)

SNA typically involves the calculation of various metrics at both individual and group level.  This is illustrated in how many of the shortest communication paths between two people pass through a third, and the visualisation of the social network as a “sociogram” . The data processing is relatively simple and there are numerous computer programs available.Trajectory Social network analysis has become increasingly popular due to the rise of the “social web”, notably twitterTM and is becoming embedded in, for example, web analytics targeted at marketing and advertising. This has been a boon Facebook as seen in its latest SEC filing.

Evolving Analytics

Netflix built its own recommendation engine, called CineMatch, and is sponsored the Netflix Prize for anyone who can improve its recommendation algorithm by 10%. Blockbuster licenses an attribute-based solution, using attributes from the a; site. Apple has a “genius” function on itunes, which employs a conventional collaborative filtering approach.

Some of the best examples of predictive analytics in practice come, not surprisingly from the hi-tech and interactive streaming media space.  Netflix awarded a $1 million prize was awarded to “Bellker Pragmatic Chaos,” a squad from AT&T Labs Statistics Research group whose system, among other things, was able to fine-tune recommendations based on “the mood of the day”—the same customer was likely to choose different movies (and rate the experience differently) on Friday evenings than Monday mornings, Blockbuster Video, and Apple’s iTunes all use various forms of recommendation tools for movies and music.

Read Part 1

More in Part 3

Written by: Catherine Casey

photo (22)

1The term “artificial intelligence” was coined by John McCarthy in 1956 and began to be applied to “expert systems” in the early 1980’s.Since then, an increasing range of new business opportunities have been discovered in the very large data-sets that are being accumulated through day-to-day business and consumer and personal activity.

One Response to “Analytics Part 2: Beyond the Zetabyte”
Check out what others are saying...

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: