Demystifying LinkedIn using probabilities

You will be able to view only 1-2 Million of profiles out of 18 Million profile on LinkedIn,

if you are a non premium customer

Since last 3 years, I spend more than half an hour on LinkedIn everyday. This is what corporate life does to us. LinkedIn is a one stop solution for almost anything you want to know in this corporate environment. I use LinkedIn for reading interesting article, knowing interesting people, marketing etc. Here is a restriction LinkedIn puts on any non-premium users :

[stextbox id=”grey”] You are not allowed to see full profile of some of the restricted 3rd level friends

(neither friends nor friend of friends) [/stextbox]

This article will give you a flavor of how you can use probability to think beyond what is visible to an average user in any social networking website. Here is simple tricks using probabilities on LinkedIn to overrule the restriction put by LinkedIn on its non premium users.

[stextbox id=”section”] Sizing of 3rd/3rd+ level locked Linkedin friend : [/stextbox]

Let’s start with estimating some numbers.

Total number of friends I have on my LinkedIn network = 1,700

Total number of 2nd degree friends on my LinkedIn network = 853,430

Total number of group users (share a group) = 973,323

The total of these numbers is the total number of profiles open to me to visit. I am likely to get profile restriction message for all other users. Let’s do a sizing of this number.

Say total number of users on LinkedIn = N

Number of friends of Mr.X = 10482

Number of common friends between Mr. X & Tavish Srivastava = 1

Say,

Probability of a person being a friend of Tavish = a = 1700/N

Probability of a person being a friend of Mr.X = b = 10482/N

Probability of a person being a friend of Mr.X & Tavish = c = 1/N

For independent  events,

a * b = c

N = 1700 * 10482 / 1 = 17.8 Million

Number of profiles not open to me = 17.8 MM – 1.8 MM = 16 MM

[stextbox id=”grey”]

~16MM profiles on LinkedIn are not open to me because LinkedIn thinks I do not know the person enough

[/stextbox]

[stextbox id=”section”]  2 steps to unlock details of 3rd/3rd+ level Linkedin friend : [/stextbox]

LinkedIn restrict users to view a profile if it thinks you don’t know the person enough. This is the screen you get

We will like to open the full profile of this person without upgrading our account to premium. Let’s try to understand what factors linkedin takes into account before deciding whether a person X knows Y.

[stextbox id=”grey”]

Here are the behavioral attributes I can think of :

1. Number of common friends between X and Y

2. Number of common communities between X and Y

3. Profile matches between X and Y (University, company)

4. Any specific information X knows about Y ( like Email ID)

5. Last page viewed by X before landing to page of Y

[/stextbox]

4 out of the 5 attributes of this list are very difficult to trigger unless until X actually knows Y. Last criterion is something one can easily manipulated. Here’s a simple way to do the same.

X reaches a Y’s profile but Y’s profile is locked. LinkedIn has a tab on left hand side called “people also viewed”. Consider following event :

P(A) : Random person views Y’s profile

P(B) : Random person views B’s profile.

B’s name comes in Y’s  “people also viewed” tab 1st name. Hence, P(B/A) is extremely high.

We all know that P(B/A) ~ P(A/B) or both move in the same direction. If P(B/A) is high P(A/B) will also be high.

[stextbox id="grey"]
Step 1 : X clicks on B's icon in the Y's "people also viewed tab 
(As illustrated in the following figure) 
[/stextbox]

Hence, if X goes into B’s profile from this page, it is highly likely that X will be on B’s “people also viewed” tab. Also because X reaches B through Y’s profile and not a random search, he triggers the 5th criterion (among the 5 attributes identified above) and will be able to see complete B’s profile.  

X finds Y on 4th number in the B’s “people also viewed” tab.

[stextbox id="grey"]
Step 2 : X clicks on Y's icon in the B's "people also viewed tab
(As illustrated in the following figure) 
[/stextbox]

This time however X triggers 5th criterion for Y as well and hence linkedin shows X Y’s complete profile.

[stextbox id=”section”] End Notes : [/stextbox]

The algorithm works in most of the cases, however, in some cases you might not find the target in the list of  “people also viewed” of the intermediate person. This is probably because you chose an intermediate person who might be much more popular than target person. In such cases try choosing people who have a closer relation to the target person. Please note that the article is based on my experience with analytics and frequent usage of LinkedIn, there is a possibility that we might have missed some of the variables LinkedIn takes into account. Do let us know of any attribute you think might be one of the variable LinkedIn takes into account and we missed to mention the same in this article.

Did you find the article useful? Share with us any other techniques which you know to make use of social networks in a better way. Also share with us any live examples of using probabilities in real world scenarios. Do let us know your thoughts about this article in the box below.

If you like what you just read & want to continue your analytics learning, subscribe to our emails or like our facebook page.

Tavish Srivastava 18 Apr 2015

Tavish Srivastava, co-founder and Chief Strategy Officer of Analytics Vidhya, is an IIT Madras graduate and a passionate data-science professional with 8+ years of diverse experience in markets including the US, India and Singapore, domains including Digital Acquisitions, Customer Servicing and Customer Management, and industry including Retail Banking, Credit Cards and Insurance. He is fascinated by the idea of artificial intelligence inspired by human intelligence and enjoys every discussion, theory or even movie related to this idea.