Your by Accenture mentioned that 94% of companies

Your everyday decisions about which
product to purchase, movie to watch and track to hear are being decided by
recommendation systems. Recommendation systems are Machine Learning
algorithms-based engines or platform that allows companies to increase customer
engagement and revenue by helping their customers find personalized services
and products that they might not be able to find by theirselves. By both
analyzing the user’s current browsing data and studying the past shopping
behavior, a recommendation system could provide high-accuracy recommendations
which would lead to customers’ needs satisfaction. For these reasons, many
companies, at least the big ones, employed recommendation systems, especially
by the ones who are in the E-commerce field.
            The amount of the number of companies
who are using MI solutions to boost their sales is growing remarkably.
According to (Framingham, 2017), the worldwide expenditure and investment on
cognitive and AI technologies will increase considerably over the next couple
of years, and that will result in generating more than $46 billion worth of
revenue by 2020. Cognitive applications, which include systems that are capable
of generating predictions, will receive the largest portion of the investments
by $4.5 billion. Another research by Accenture mentioned that 94% of companies
agreed to the fact that personalizing their services or products is essential
for current and future business success (, 2015). For this reason,
big companies such as Amazon, Netflix and Spotify were among the first
businesses in their industry who believed on this trend and invested the money
and time on it. In fact, the main businesses of these three companies are fully
relying on their recommendation systems. In this essay, I will provide an
overview of how Amazon, Netflix and Spotify employ the recommendation systems
and what benefits they reaped.
is the first real-life example of an E-commerce company who is using
recommendation systems match their customers’ needs and boost their sales. 35%
of Amazon’s revenue is because of its recommendation engines (Yash, 2017).
Amazon treats their recommendation system as their main targeted marketing tool
in both their online store and email campaigns. Last year, Amazon has given
away to the public it’s Deep Scalable Sparse Tensor Network Engine (DSSTNE)
which is a powerful AI framework and algorithms that Amazon has been using to
help to fully reach their customers’ needs and expectations for long years
(Klint, 2016). Many specialized websites and interested individuals analyzed
this sophisticated framework to reveal the secret of Amazon’s personalized
digital experienced success. According to Tom, There are different ways Amazon
use for the on-site recommendations and off-site recommendations (Email
            With regards to the on-site recommendation
for Amazon, every consumer’s Cart represents a real payment process for real
customers. When costumers brows the online store and click on the
recommendation link, they will be directed to a page that has many products
recommended just for them and that customers are likely going to check them
out. One recommendations list is the ” Frequently Bought Together” which its important
aim is to raise the overall orders. Another one is the ” Your Recently Items
and Featured Recommendations”. The company view what kind of products the
customers have viewed previously during their online set and then suggest a
variety of products that are alike to the ones they browsed but with many
different options such as; in different colors, sizes, to increase the chance
that the customer will at least find one product that he cannot refuse. A
similar recommendation category is the “Related to items you’ve viewed” which
works almost the same way. Amazon’s philosophy on their ” Your Browsing
History”  category is that if a customer browsed a specific products
earlier that means he has an interest in the products, so they put it out there
in front of their eyes to remind their them of interests. Tom thinks that the
goal of the  ” Customers Who Bought This Item Also Bought” recommendation
category on Amazon website is just a technique from Amazon to encourage selling
products that don’t have enough popularity. In that way, Amazon also will help
the retailers to increase the moving inventory. According to Dokyun and Kartik,
the ” Customers Who Bought This Item Also Bought” category increased the number
of product’s views by 25% and the number of products purchased by 35%. The
“Best-selling” list provides the customers with the variety of top sellers. The
list of best sellers of an exact product might increase the probability of the
customer viewing and maybe buy popular products from different categories that
they might haven’t the will to buy before. The last recommendation list of the
on-site recommendations, is a recommendation that I was impressed of how smart
is it, which is “There is a newer version of this item” list. Amazon takes
advantage of how people love to own the new versions of their loved kind of
smart phones and other electronic devices.  When users view their previous
invoices of any electronic devices that they have bought in the past, they will
find a recommendation next to it which alert the customer about a new version
of the electronic device is available in the store and waiting for your magic
simple click.
            Amazon’s off-site recommendations
through the email are not less important than the on-site recommendations. One
way of Amazon recommending products to their customer’s trough email is by
sending an email that shows the customer with the range of best selling
versions of a specific items category they purchased or even browsed in the
past. For example, if a customer on one on-site set of shopping were only
browsing for Samsung Android phone, he/she will probably receive an email that
has only a variety of different models or versions of best selling Samsung
Androids. Another email the same customer might receive, will have products
that frequently purchased together, and in case of the customer who was looking
for a Samsung Android, the email may contain accessories that are related to
the main product. The last type of email that Amazon sends to their customers
to encourage them to end up buying a product is the most selling products from
the specific product category that the customer was browsing. In this case, the
brands don’t matter. For the customer who was brewing Samsung smart phones,
he/she might receive an email that contains a range of popular smart phones
from Samsung and other sellers.  

            Do you know that more than 80 % of
the shows that have been watching on Netflix by users were suggested and
exposed through their recommendation system ( Libby, 2017). In explaining how
the previous fact is an interesting, Libby says that “That means the majority
of what you decide to watch on Netflix is the result of decisions made by a
mysterious, black box of an algorithm. Intrigued?,,”. Netflix uses different
algorithms to analyze a variety of users’ activities. Data collected such as;
what movies or shows users searched, played, or rated with the date and time.
They also collect data about users’ scrolling and browsing behaviors. All these
data are being analyzed by several algorithms each designed for different
purposes, and then it resulted in providing their users with recommendations
that match their preferences in the TV shows and movies. Furthermore, this
information gets combined with other data that its purpose is to understand the
TV shows and movies’ contents. Moreover, Netflix has a team of in-house and
freelancers who watch every show on Netflix and tag each one to different
detailed attributes For example, whether there was a corrupted cop in the show.
The later, these tags and the users’ behavior data are being analyzed by using
advanced algorithms. Finally, the viewers categorized into thousands of
different taste groups. The result is that the recommendation will be improved
to match the viewer taste. For example, the types of rows that will display to
the viewer and how it is ordered. Also, the recommendations will take into
consideration the language, country and cultural context for each user (Libby,
2017). According to Netflix executives Carlos and Gomez, the company managed to
save $1 billion each year because of their recommendation system, and this
enables them to use this savings to invest in new shows that the users will
continue to watch and thus allow them to maintain excellent Return on
Investment (ROI). 
            Besides Amazon and Netflix as two
examples of E-commerce companies who are considered to be the first who smartly
started developing and using recommendation systems to personalize their
customers’ experience and increase their profits, music recommendations are
really new, and Spotify, a music, and video streaming app, is the only one, at
least perfectly, who developed a sophisticated and  complex machine
learning framework and led the competition over its traditional competitors
like Apple and Pandora.  Spotify uses machine learning to create an
algorithm-based playlist that matches their users’ taste, and put out there into
the Discover Weekly in weekly bases. According to Sophia (2010),  Spotify
has a unique way of creating their recommendation engine (Discovery engine).
They combine and mix the three well-known recommendation models that used by
other companies. The three models are the Collaborative Filtering, Natural
Language Processing (NLP) and Audio models. 
Spotify employs the Collaborative Filtering model in a way
that slightly different than the other companies who use the same model such as
Netflix. Not similar to Netflix, who uses the rating stars from their users as
a guide for them to develop a recommendation engine for other alike users,
Spotify takes in consideration the stream counts of the songs, as well as data
includes whether the user add the song to his favorite list, or viewed the
singer’s page. An example to explain how the previous model works would be two
users (Naif and Mohammed) who share the same preferences for 2 tracks (track A
and B, respectively) and each one of them has favorite anther track that the
other user has not (Naif likes track C, and Mohammed likes track D). Therefore,
Spotify predicts that both users probably have similar taste and they are
likely going to enjoy the songs that they don’t listen to, but the other did.
So, Spotify in this case, will suggest Song C for Mohammed and D for Naif). 
            The second recommendation model which
is the NLP used by Spotify to analyze the text through the web. This model
searches through the internet to find blogs and other texts that are about
music. Then it collects data about what people say and think of some songs and
artists. By collecting this data, they could figure out which songs are similar
to others and this boosts their accuracy in matching their users’ taste.
             The Audio model is the third
one used by Spotify, and it is the most complex one. They use convolutional
neural networks to analyze audio data. convolutional neural networks analyze
the songs to understand the songs and its different characteristics such as the
loudness, mode, and more other characteristics. The understanding of such a
complex details and characteristics of a song, enable Spotify to find and fully
understand the similarity between multiple songs. As a result, the incredible
accuracy of matching users’ taste on songs will be reached which, therefore,
increase the user’s loyalty to the service. 
            To conclude, the huge benefits of
employing MI recommendation systems models by E-commerce companies such as
Amazon, Netflix, and Spotify worth the effort and the money. Besides the
increasing of revenue as a benefit of using algorithms-based recommendation
systems by these companies, the value the companies deliver to their customers
will let them assure that they gain their customers’ loyalty which is essential
for the companies to stand out in competition. According to Mark, Loyal
customers account for 25% to 40% of the total revenue for companies that under
the SumAll group, SumALl is marketing analytics company with over 200K clients.
Lastly, companies that have loyal customers accounted for 40% result in almost
50% of more revenue than other companies that only have 10% of loyal customers. 

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

I'm Harold!

Would you like to get a custom essay? How about receiving a customized one?

Check it out