ITM​ - ​Final questions 

Short answer 

Big Data​ = High-volume, massive almost always unstructured text data (90%+). 
 
Enterprise Data Warehouses​ = Data warehouses that pull together data from disparate sources 
and databases. 
 
Classification performance measures​ (accuracy, precision, recall, and F-measure) = 
Accuracy = (TP + TN)/(TP + TN + FP + FN)  
Precision = TP/(TP+FP)   
Recall = TP/(TP+FN)  
F-measure = 2 x [(precision x recall)/(precision + recall] 
 
Base-rate fallacy​ = P(Terrorist | Alarm) → p ​ attern of reasoning rendered invalid by a flaw in its 
logical structure.  
 
Clustering​ = Users make groups of data to determine patterns from the data. 
 
Social media​ = Platform to engage with customers and deal with their issues in real time 
increasing quality of service (QoS) and customer trust. 
 
Social network service​ = p​ latform to build social networks or social relations among people who 
share similar personal and career interests, activities, backgrounds or real-life connections. 
 
Private SNS​ = More private, personal SNS. 
 
Cross-channel​ =​ Customers see multiple, but separate channels with which they can interact with 
the same brand. Retailers have a single view of the consumer, but continue to operate separate 
channels as silos. 
 
Same as multichannel, but with an added twist: instead of hitting your target with the same 
message across multiple channels, you use different messages that are strung together 
throughout your different channels.   
 
Omni-channel strategy​ = Customers experience based on brand, not channel within a brand. 
Retailers leverage the functionality of each channel as well as their single view of the customer to 
provide optimal overall experience. 
 
Multichannel approach to sales that seeks to provide the customer with a seamless shopping 
experience whether the customer is shopping online from a desktop or mobile device, by telephone 
or in a bricks and mortar store. 
 
QR code​ = ​QR code is an alternative to typing a URL address into a mobile browser.  
 
Sponsored ads ​= T ​ o have their ad appear in Google sponsored links, companies can enter an 
auction to show their ad if it's relevant to a user's search query. This is called pay-per-click 
marketing, because the advertisers pay Google for clicks on their ads. 
 
Display ads​ = a
​ large advertisement, especially in a newspaper or magazine, that features
eye-catching type or illustrations.

SML​ (Social Media Listening) = P

Google Analytics​ = M​ easure your advertising ROI as well as track your Flash, video, and social 
networking sites and applications. 
 
Benevolence​ = act of kindness? Wtf? 
 
Credibility​ = Reliability of the information?

Customer review​ = ​Review of a product or service made by a customer who has purchased the
product or service. Customer reviews are a form of customer feedback on electronic commerce and
online shopping sites. There are also dedicated review sites, some of which use customer reviews as
well as or instead of professional reviews. The reviews may themselves be graded for usefulness or
accuracy by other users.

Rating (volume and valence) ​=  

A. Business executives asked ”Isn’t big data just another way of saying analytics?”. Please 
explain how big data is different from business analytics in 3 respects. 
 
Organizations around the world and across various industries are employing big data in order to 
gain more insight into their business and improve processes. 
Big data is different from business analytics in 3 key aspects: 
Volume​: Large volume of structured and unstructured data are analyzed. Consider that 90 
per cent of the data today was created only in the last two years. 
Variety​: The analytic environment has expanded from pulling data from enterprise systems 
to include big data and unstructured sources. 
Velocity​: Speed of access to reports that are drawn from data defines the difference 
between effective and ineffective analytics. Real-time or near real-time information access enables 
organizations to be quicker in making decisions and executing moves than their competitors. 
Big data may be unstructured and unwieldy, but there’s an enormous amount of vital signals 
among the “noise” that accompanies it. The value is just waiting to be mined. 
 
B. Explain the rise and fall of Tesco PLC. And that big data cannot always be a panacea for 
firms. 
 
Tesco  PLC  has  7,600  shops  in  12  countries  under  different  brands  and,  in  2014,  $110  billion  in 
group sales and $3.6 billion in profit. 
-  The  company  introduced  ​Clubcard  Loyalty  Program  in  1995  (38  million  card  members  -  16 
millions  active).  This  loyalty  program  informs  the  company  about  the  content  of  shopping  carts 
and  thus  what  are  the  best  products,  which  product  combinaisons  are  interesting, how should the 
shop  be  organized  to  maximize  the  customer  willingness  and  temptation  to  buy  (e.g. 
beer+diapers). 
 
-  Link  local  weather  forecast  with  18  million  product  items and local store information and directly 
forwarded to suppliers so it reduces inventory and food waste. 
 
-  The  Analysis  of  refrigerator  data  with  IBM  research  laboratories  reduced  $25  million  a  year  of 
energy bill by optimizing in-store refrigerators. 
 
However, the company failed because of:   
-  Rich  and  data-rich  loyalty  programs  and  analytics  capabilities  can’t  stave  off  the  competitive 
advantage of slightly lower prices and a simpler shopping experience. 
- Loyalty cards are more helpful to the supermarket than they are to the shopper. 
- Failed to globalization. 
 
C. Explain how classification techniques and clustering techniques are different with a real life 
example - e.g., for what types of questions, what methods should be applied. 
 
Both classification and clustering​ split data into groups.​ However, ​classification is s ​ upervised 
learning​ and clustering is ​unsupervised learning​. 
 
With c ​ lassification​ the groups (or classes) are specified ​before hand​, with each training data 
instance belonging to a particular class. It is the a ​ ssociation​ between the instances features 
and the class they belong to that classification algorithms are supposed to learn. An example 
of use would be an insurance company trying to assign customers into high risk and low risk 
categories. 
 
With c ​ lustering​ the groups (or clusters) are based on the s ​ imilarities of data instances​ to each 
other. No predefined output class is used in training and the clustering algorithm is supposed 
to learn the grouping. An example of use would be an online movie company recommending 
you a movie because other customers who had made similar movie choices as you in the past 
(i,e. the cluster you fall in) have favorably rated that movie. 
 
D. Explain why tightening filtering-out criteria of passengers in an airport does not necessarily 
result in higher level of security in the airport in 2 respects. 
 
Dessiner le graphique: Increases risks of false alarms. 
 
E. Single out an example of failure in utilizing social media and describe details. After then, 
please suggest a better way to overcome the problem. 
 
One example of failure in utilizing social media was done by JPMorgan Chase. After incurring more 
than 5.8 billions trade losses in the credit default swaps market in 2012, the company posted a 
tweet asking people to ask them career advices. Obviously, angry customers used the opportunity 
to complain and bash the company.   
You have to quickly respond to these kind of negative comments and try to prevent them from 
spreading. One solution can be to delete the controversial comment/post and apologize publicly. 
 
F. Even if Xiaomi recently has a slump in smartphone sales, it has shown surprising growths. 
Describe 3 pillars to Xiaomi’s success. 
 
1​) Smartphones featuring high-end specification, yet coming at prices significantly cheaper than 
their competitors. 
2​) Strategy of selling directly to consumers online, making particular use of flash sales as a way of 
generating buzz and scarcity. 
3​) Strong social media presence, which it has used in lieu of traditional advertising, again a 
cost-effective strategy. This social media focus has fed into product development, with the 
company incorporating feedback from customers into future updates, further strengthening the 
bond with consumers, and giving rise to so-called ‘Mi Fans’. 
  
G. Academicians and practitioners claim that marketers develop relationships with bloggers, 
encouraging them to write positive stories about the brand and even employees to post 
positive opinions. Describe why the strategy does not work well in reality and suggest better 
way to manage the blogs. 
 
Customers usually don’t like watching/reading sponsored reviews of products because they 
believe they are not 100% honest as the person is being paid for it. Thus, the trust in the quality of 
the product diminish and also in the honesty of the person reviewing it. Same with employees, it 
seems obvious that they are not going to say anything negative about the products their company 
is selling by fear to be fired or have negative impact on their careers. “Forced positive reviews” 
have no value to the average consumer because they are considered untrustworthy.  
The company can create its own blog and define and understand its target audience, then provide 
and adapted content and finally track, measure and adapt their content to the received feedback. 
 
H. Explain omni-channel strategy and compare it with cross-channel strategy. In addition, 
describe a company’s example on how successful the company was. 
 
Omnichannel  is  a  multichannel  approach  to  sales  that  seeks  to  provide  the  customer  with  a 
seamless  shopping  experience  whether  the  customer  is shopping online from a desktop or mobile 
device, by telephone or in a bricks and mortar store.  
What  distinguishes  the  omnichannel  customer  experience  from  the  multichannel  customer 
experience  is  that  there is true integration between channels on the back end. For example, when a 
store  has  implemented  an  omnichannel approach, the customer service representative in the store 
will  be  able  to  immediately  reference  the  customer's  previous  purchases  and  preferences  just  as 
easily  as  the  customer  service  representative  on  the  phone  can  or  the  customer  service  webchat 
representative  can.  Or  the  customer  can  use  a  desktop  computer  to  check  inventory  by  store  on 
the  company's  website,  purchase  the  item  later  on  with  a  smartphone  or  tablet  and  pick  up  the 
product at the customer's chosen location. 
 
I. Describe how different mobile internet is to PC-based internet with your speculation. 
 
Mobile Internet is somewhat less “Internet-like” -> Not the same sensation, less information, no 
“clicking”...  
For PC users, moving one position upward in rank yields an increase in the odds of clicking on that 
brand post by 25% 
For PC users, a one mile decrease in distance between a user and a brand store yields an increase 
in the odds of clicking on that brand post by 12% 
 
But, interestingly 
 
For mobile users, moving one position upward in rank yields an increase by 37%. 
For mobile users, a one mile decrease in distance yields an increase by 23%.  
 
J. Describe characteristics of cosmetic products and two focusing strategies of Sephora to 
utilize the characteristics. 
 
When it comes to cosmetic products, customers’ need are diverse and can change according to 
life events. Cosmetics are experienced goods and one does not fit all, thus exploration and 
experimentation are very important. 
Sephora then decided to provide an environment where customers could try m ​ ultiple brands​ in an 
“​assisted self-service​”. 
 
K. Describe strategies of [Link] in 4 stages (Trigger, Consider, Purchase, and After 
purchase). To begin with, [Link] offered sophisticated search functionnality along 
with details about every product the company sold in order to help customers find what her 
friends or celebrities used or find what she has used. 
 
Trigger​: If you have clear needs, you can go [Link] directly to get them easily.  
Consider​: [Link] offered sophisticated search functionality along with details about every 
product the company sold.  
Purchase​: [Link] offers free shipping for orders over $50, and three free samples with every 
order.  
After purchase​: [Link] keeps a forum where customers could share, discover, and learn. 
They also enabled customers to post comments and reviews on their products. 
 
L. Describe the logic behind that RIM was not successful in the competition with iOS and 
Android. 
 
Despite its early success before 2008, RIM ended up losing sales and its market share when iOS 
and Android phones were released. Its main errors were to fail to sustain innovation and to 
consider a broader market. RIM wasn’t fast enough for the ​technology race today.