Application of Social Network Analysis: Case Study
Let's take this page for example,
i.e. https://www.facebook.com/USAiPhone5S. This
is the official page of iPhone5S, and will be the object of our "Social Network Analysis".
First and foremost, it is important to define the objective for carrying out this Analysis:
To help take the “voice” of
the potential and existing customers, i.e. users on the above Facebook page, into various parts of the organization, to enable organizations and their departments to be able to“hear” the voice of users in different
ways, to help organization(s) understand the AIOs (attitude, interest and opinion) of various users
towards iPhone5S as an aggregate and as various AIO clusters. The study should also tell us what product
features, attributes, news etc. certain users like and and how these users are
connected to each other through various similarities.
Among other business insights, this would help with the following:
1> Releasing pertinent
information at Facebook which is likely to be of interest to the largest number
of users so that they remain engaged through the web-site, i.e. to keep the
user on the AIDAS path
2> Understanding the geographic co-ordinates of various users and
their interests, so that on-the-ground action can form a basis
3> Gather information from users how they compare the product and its
various attributes to competitor and their products – so that those inputs can provide evolutionary direction and insights
4> Overall answer a huge number of strategy questions – including
forecasting the fate of the product in terms of sales response that can be
expected, to training needs of salespersons to address most expressed concerns
and overall ability of the company to maintain sustained interest in its
products and offerings
Data Definition / Data Attributes of GDF file:
name VARCHAR, -> field is ‘name’ and data is of the type
VARCHAR (variable character data)
label VARCHAR, -> field is ‘label’ and data is of the type
VARCHAR (variable character data)
type VARCHAR, -> field is ‘type’ and data is of the type
VARCHAR (variable character data)
type_post VARCHAR, -> field is ‘type_post’ and data is of the type
VARCHAR (variable character data)
post_published VARCHAR,
-> field is ‘post_published’ and data
is of the type VARCHAR (variable character data)
post_published_unix INT,
-> field is ‘post_published_unix’ and
data is of the type INT (integer)
user_locale VARCHAR,
-> field is ‘user_locale’ and data is
of the type VARCHAR (variable character data)
sex VARCHAR, -> field is ‘sex’ and data is of the type
VARCHAR (variable character data)
likes INT, -> field is ‘likes’ and data is of the type INT
(integer data)
likes_count_fb INT,
-> field is ‘likes_count’ and data is
of the type INT (integer data)
comments_all INT,
-> field is ‘comments_all’ and data
is of the type INT (integer data)
comments_base INT, -> field is ‘comments_base’ and data is of the
type INT (integer data)
comments_replies INT, -> field is ‘comments_replies’ and data is of
the type INT (integer data)
comment_likes INT, -> field is ‘comment_likes’ and data is of the
type INT (integer data)
shares INT, -> field is ‘shares’ and data is of the type INT
(integer data)
engagement INT, -> field is ‘engagement’ and data is of the type
INT (integer data)
post_id VARCHAR, -> field
is ‘post_id’ and data is of the type VARCHAR (variable character data)
post_link VARCHAR ->
field is ‘post_link’ and data is of the type VARCHAR (variable character data)
Key Findings:
·
The most recent photograph had so far been liked
by 63 members. The speed with which a a
particular post gains popularity or liking on time-scale can determine nature
of future business communication
·
Users are distributed geographically into
clusters, as they come from different countries
· Gephi enables visualization of users in various terms, including geographical
clusters
·
For example, 845 people liked the photograph of golden
iphone in one post, 882 in another. User
level mining of such “liking” clusters may yield the unique number of users per
geography that are interested in the golden iphone. When data is aggregated and then divvied up
based on these and other parameters, it can point to a certain possible
e-demand, i.e. these net savvy users may
be served with (let’s say Apple’s) business communication which may provide
links to these users from where to get more information and even from where to
buy the product. With this information,
basically the company would have executed AIDAS model (awareness -> interest
-> demand -> action -> sale)
·
Therefore, from business perspective it is not
only sufficient to provide information but also to lead the customer through the
AIDAS model through implementation of 7P’s (product, price, place, position,
packaging, people, processes) so that awareness to sale cycle can be effected
through e-and physical presence
·
People from various countries, geographies,
gender, education background etc. are converging to like iphone golden cover
·
This shows that iPhone with golden cover has not
only been widely accepted, but also has set an altogether new trend for gold
colored devices
·
Users can also been viewed in terms clusters of
likes for various attributes of the iPhone
·
It appears that more in some threads one gender
is more interested than the other. Such
threads can be analyzed to understand gender sensitivities and preferences,
e.g. for pink iphone etc. such inputs can then be used for new product design (
a more compact iphone for usually smaller hands of women, even lighter at
times) or for offering mere cosmetic additions (like a pink cover) etc.
·
While likes may be common, even then many of the
comments relate to different business interest areas – for example some users
are asking when will the product come to their country, others want to know the
price, will the product be available unlocked, etc. etc.. This means further mining is necessary
between likes and comments – to see how many likes and how many comments of the
same or similar type are there, so that these can be sent to the relevant
business department of Apple for being addressed in a proper way
·
Country level data can also tell us through
comments some of the problems with iphone, for example a 5.0 inches iphone may
be too big for the Japanese customer. So
the company may decide to acknowledge the Japanese comments/customers and
design product accordingly
·
Pricing information from various customers and
clusters would tell the company whether it has the right pricing policy for
that country or customer group/cluster.
Apple may however still decide to keep premium pricing even though users
in a cluster comment that it is too expensive – so as to keep its image
·
Product usage and support related comments may
again help design better product(battery running out too fast complaints would
normally come from a region where electricity network is not that reliable yet
or if there are not sufficient public charging options – that can lead Apple to
devise partnerships with companies to provide private charging stations at
public places) or give rise to new training needs, not only for salespersons
but also for customers so that they know how to use the product properly. Analysis of clusters can tell us that
·
Thus, Apple may also be able to identify need
for new partnerships in areas where it does not have a competence, e.g. with
companies that offer public phone charging points
·
Most recent posts (downloaded 10 recent posts
data at around 10:40 AM on November 8, 2013) have posts and comments from as
far as 6th of November. The
exact hour is also available. This shows
the echoing impact of latest posts – these seem to be echoing over 3 days. This can be used by the marketing department
of Apple to design messages so that they can reasonably echo via comments for
sufficient time so that the marketing or any service or other message gets
maximum exposure
·
The average life of a post can be determined for
categories of messages which can again help in optimizing communication from
Apple
· Many of the comments are in languages other than
English. This has an implication for
Apple and those interested in interpreting comments for business – that adds
requirement of multi-lingual personnel in various departments because as
already stated, comments are related to various topics and therefore to various
departments – so many departments may need to have multi-lingual personnel or
as an alternate strategy, a global multi-lingual desk may be operated by a
company to translate all comments into English and then be distributed to
within-Apple departments for their contemplation and further strategization, or
a country-language-wise desk may be created (as an alternate strategy) based on
how many people are interested from which part of the world. So there can be various strategies and
information will help uncover the right strategy
· A detailed analysis can also reveal the distribution of language(s) over geography - for example to answer the question, in which state of the US more stores should have bi-lingual salespersons?
Where should Apple have what kind of
distribution? Where should Apple have
more stores or just be content with e-presence?
Take example of US versus India.
Since a lot of posts were in the past from the US, Apple opened more
distribution centers there, but had more e-presence in India. Now Apple has caught frenzy in India also, so
Apple is now expanding distributor network here also. Similar approach can help uncover other
emerging markets
·
A number of people have liked various attributes
of the iPhone and those attributes are showing various growing trends,
indicating strong support for the iPhone on all those attributes
Alright - so much for now, more later. If you have some thoughts, please feel free to send them in. Eager to hear!
No comments:
Post a Comment