Week 13: Clear

Hunt versus Hike -

The differences in Facebook’s & Google’s ad models

 

Have you ever heard of hunter and gather? Probably yes.

But have you also heard of the hunt versus the hike? No, it does not have to do with any leisure activity or hobby. It is about the different ways users or costumers are using websites. Moreover this difference has a great influence on the ad models and in turn on the revenue models of Facebook and Google.

 

Graphic 1&2: Hunt vs. Hike

  

People using search engines like Google are on a hunt.

A hunt is described as “a task-oriented search to collect information that will drive a specific action”. Users want to get specific information to learn something, they want to get questions answered, they are searching for a solution to a problem or are just looking for something they want to buy. Google and other search engines makes use of the fact that the hunt might overlap with ads. So if you are for example searching for a trip to Europe, Google will show advertisements of travel-related companies like hotels, rental cars, travel providers, or if you want to look for a nice dress, Google will show advertisement of online shops offering dresses and other clothes. The advertisement is kind of targeted to the search terms you use. Hence, you are on a hunt and Google provides potential prey.

 

A hike refers to the activity when you have a vague or rough idea what you will encounter. However, you are not looking for something specific but you rather want to have a look around or explore. Facebook is one website which offers a great opportunity and possibility to go on a hike. You just click on a news feed of a friend, which brings you to photos of a recent party, on which people are tagged. You click on the tag of a nice guy you met and explore his Facebook profile. Then you see his favorite music artists and you click on the group page of a singer. This page is linked to a music festival you explore. As you think your close friend might be interested you share this public event with her and so forth. When being on a hike you kind of walk from one thing to the next without a specific planned route.

 

Google’s revenue model is based on the concept that it charges text advertisers if and only if customers click on their advertisements. Without any clicks, the firm advertises for free. Hence, with respect to ROI (return on investment) it is very efficient. Google users tend to click on advertisement ca 2% of the time.

Facebook also tried to leverage on offering space for online advertisement. It charges companies CPM, so each time when the advertisement was shown to a customer, Facebook charged the advertisers, and the payment rate was calculated per 1000 impressions. However, Facebook achieved only a click-through rate of 0.04% and it pulled its banner ads in in early 2010.

Graphic 1 by Florian Prischl on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Hiking_boots_on_sand.jpg

Graphic 2 by Fantagu on Wikimedia.org:

http://upload.wikimedia.org/wikipedia/commons/4/49/Modern_Hunting_Rifle.jpg

Week 12: Clear

Domain Name Service

Have you ever wondered about how it works that when you type in a name or link of a website into the search field of your internet browser, the website you wanted to look at, turns up?

The answer to this question is with the help of Domain name service or shortly called DNS. The DNS is a database that looks up the host and domain names that are entered and returns the actual IP address. You can think of it like a huge collection of phone books being called nameservers which are able to find e-mail servers and web-servers.

Graphic 1: Illustration of how DNS functions

 

Let’s look at an example to get the real underlying sense and procedure.

Imagine you want to surf on www.google.com. First of all you type this address into the search field of your browser. The DNS resolver helps your computer to find this website address, as your computer alone is not able to find it. The DNS resolver looks into the DNS (the “phonebooks”) listing host/domain name combinations, to find the matching IP address. However, it alone is not able to find the IP address, so it contacts root nameserver. These root nameservers cannot provide the IP address, but they are a good starting place for searching, as they will point to a .com nameserver, because of the “.com” ending. These .com nameserver won’t know the right IP-address itself, but will point at a google.com nameserver. Finally, the google.com nameserver knows the IP address and can respond to the query of the DNS resolver. Then the DNS resolver receives the IP address from the google.com nameserver and will provide that information to your computer.  As soon as your computer knows the IP address it will be able to communicate directly with www.google.com and you are able to serve on www.google.com.

Whenever you have visited one website your computer stores that information, so the IP-addresses, in a storage called “cache”. The cache makes it possible to speed up your searches if you search for websites you have already been to. You can think of it as bookmarks in the telephone books. You know where to look for the information much faster. So the whole procedure explained above just takes place if you visit a website for the first time.

Original graphic by Everaldo Coelho on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Computer_n_screen.svg

Original graphic by George Shuklin on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Server-tower.svg

Original graphic by GNOME icon artists on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Gnome-network-server.svg

Modified by Nina Maria Scherl.

Week 11: Clear

Data Mining

Every day companies create heaps of data which has to be stored, processed and analyzed, and the total amount skyrockets. Thanks to Moore’s Law, which holds true so far, the prices for storages and processing capacity decrease, making the storing and processing of such vast amounts of data possible and affordable. However, modern data sets can be so large that it might be impossible for humans to recognize and spot the underlying trends. Stated differently, a good analysis done by individuals is difficult if not even impossible.

Graphic 1: Huge Amount of Data

 So how can we analyze these data in an efficient and effective way?

Which tools can help us doing it?

For the answers to these questions data mining comes into play. Data mining is the process of using computers to identify hidden patterns, and to build models from large data sets. With the help of computers and automated searches, people are able to analyze data and get sufficient and helpful insights and results.

There are many areas where data mining can be used to analyze data. Some of the key areas are:

Graphic 2: Areas for Data Mining

Hiring and promotion to determine characteristics which are related and consistent with employee success.

Fraud detection to uncover patterns being consistent with criminal activity.

Marketing and promotion targeting to examine which customers respond to which type of offer for which price and at what time.

Customer churn to find out which customers are likely to leave, and create and employ strategies and tactics to retain these customers.

Market basket analysis to identify which kind of products customers are likely to buy together or at the same time, in order to enable the company to cross-sell more of its products and services.

Collaborative filtering to make an experience of an individual customer personalized by basing it on the preferences and trends examines across similar customers.

Financial modeling to build trading systems to capitalize on historical trends.

Customer segmentation to analyze and find out which customers are most valuable to a firm.

Data mining seems to be quite helpful and usable in many areas. However, for data mining to work two critical conditions have to be met. 

First of all the organization must have clean, well-organized, and consistent data, and second, the events in that data should reflect current and future trends in order to analyze and predict more reliable trends.

Graphic 1:

Original photo by gruppetto on wikimedia.org:

http://commons.wikimedia.org/wiki/File:Exquisite-folder_pictures.png

Original photo by Przykuta on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Exquisite-folder_document_vip.png

Original photo by BJJ_White_Belt.svg on Wikimedia.org:

http://commons.wikimedia.org/wiki/File:Black_belt.svg

Modified by Nina Maria Scherl.

Graphic 2:

Original photo by Angie on wikimeida.org:

http://commons.wikimedia.org/wiki/File:Shovels.jpg

Modified by Nina Maria Scherl.