1. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
HIGH NOTE RECOMMENDATIONS
EXECUTIVE SUMMARY
Recommendations
To increase adoption to premium (paid) service, High Note should focus on the following strategies in
order of prioritisation, based on potential ease and cost of implementation:
1) Expand difference between free and premium service, and ensure that users fully understand
the benefits of the paid service
Having an expanded difference between free and premium service will encourage more free
users to convert to premium service to enjoy the additional features. Specific exclusive features
for premium users could include exclusive access to the latest songs, offline listening (like
Spotify), and unlimited song listening.
Importantly, the exclusive features for premium users should be clearly explained and
understood so that free users will be encouraged to convert to premium service. This can be
done through in-app reminders, advertisements, direct email, etc. A premium service free-trial
period could be introduced so that free users can experience the different service level and be
enticed to convert.
2) Encourage premium users to add friends to their network
Premium users influence free users to convert to premium service. Therefore High Note should
introduce efforts to connect premium and free users. Strategies could include connecting
premium and free users (i.e. recommending friends) based on similar taste in music (using data
on their songs listened and playlists created), introducing a feature for users to share playlists
and thereafter connecting creators and users of playlists to become friends, rewarding users for
having more friends. This feature will also encourage users to create more playlists.
3) Increase the number of free users by leveraging referral and social app marketing
Focus on building a strong, active community to leverage word-of-mouth marketing. High Note
can then work towards achieving a moderate conversion rate from a high volume of traffic. High
Note could introduce a referral program to incentivize existing users to bring more users
onboard. High Note could also partner with social media platforms such as Facebook to market
the app, as well as make use of social app marketing tools to enhance High Note’s visibility
amongst streaming music users. These efforts can then be coupled with strategies to convert
free users to premium users to increase the conversion rate.
2. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
4) Encourage users to create more playlists
Users with more playlists are more likely to be premium users. High Note could introduce
features to make it easier to create playlists, to adapt playlists shared by friends, popular playlist
ranking for different playlist categories (jogging, meditation, driving, partying etc.)
Findings
Qualitative Analysis
Research on the freemium and music streaming industry, including competitors, yield the following
insights:
1) Expand difference between free and paid service. One of the chief purposes of freemium is to
attract new users. If you're generating lots of traffic but few people are paying to upgrade, you may
have the opposite problem: Your free offerings are too rich, and it's time to cut back.
2) Ensure that customers fully understand the premium offer. It could probably monetize more users
if the distinctions between free and paid offerings are clearer.
3) Increase in free users eventually leads to larger paid users. Freemium’s core growth engine is social
proof aka word-of-mouth marketing. All other things being equal, it would be better to convert 5%
of 2 million monthly visitors, for example, than to convert 50% of 100,000 visitors. The best long-
term strategy is generally to aim for a moderate conversion rate (research has shown that most
companies' range from 2% to 5%) coupled with a high volume of traffic.
4) Referral & Social App marketing. Referral marketing enables marketers to get value for acquisition
and conversion, while rewarding existing customers and generating new customers. Social app
marketing, which is relevant to music streaming, performs well in terms of user quality, conversion
and volume.
Quantitative Analysis
Regression analysis was carried out on the characteristics and behaviours of existing High Note users
showed that conversion to premium service was associated with the following variables:
1) No. of subscriber friend. Conversion to premium service is associated with the number of
subscribers friends a user has (if a user has more friends who are subscribers, the user is more likely
to be a premium user)
2) No. of playlists. Conversion to premium service is associated with the number of playlist a user has
(if a user has more playlists, the user is more likely to be a premium user)
The quantitative analysis, however, only showed association between conversion to premium service
and the variables. Therefore further analysis and/or experimentation should be carried out to show
causality.
3. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
Freemium
The word freemium is a combination of the words free and premium. It describes a business model in which you
give a core product away for free to a large group of users and sell premium products to a smaller fraction of this
user base. Freemium is not the same as a free trial. A common misconception is that freemium includes or consists
of a free trial period though this is not the case. The difference is that a free trial allows a user to test a product or
service for a limited time before buying the actual product. A freemium model on the other hand provides a product
or service that is always free and the premium part comes from offering the free platform of users additional
products or services that for a fee can expand or improve their experience. So just because a free product appears in
a business model does not make it freemium.
Product Characteristics. Product can have 2-step value proposition. The first proposition is a dead-simple single
user experience, ideally in the form of a mobile app. The second proposition is a network effect proposition or multi-
player experience. This is key to fast growth and high retention.
Market Characteristics. The market offers either tens of millions of users at low price points or millions of users at
higher price points. Freemium funnels typically convert only 2 to 4% of users who try the product.
Distribution Tactics. Target segment of users has a strong, active community. Freemium’s core growth engine is
social proof aka word-of-mouth marketing. To drive word of mouth marketing, a freemium business needs a
community, typically an existing one which falls in love with the business’s product. Harnessing this community
starts with building a great product. Then, a startup must cultivate and grow their community. Second, hackable
distribution channels are a sine qua non. Freemium business must either leverage new distribution platforms or
develop novel tactics on existing distribution platforms.
Conversion Point. The conversion to paid point feels natural because a user derives enough value from the product
that living without it is too painful. Successful freemium products have a conversion point to paid at the point the
customer understands the value proposition and is already committed to the product. I think there are 4 ways to
trigger conversion to paid:
1. Limited time trial: user pays after 14 days or 30 days after enough time has passed to kick the tires on the
product.
2. Money back guarantee: user pays on sign up but can cancel at any time.
3. Utility upsell: user pays after a certain amount of data has been uploaded into the system, or a certain
number of seats in a given company have been activated.
4. Limited functionality: user pays to integrate into other systems or to access pro-level features like exporting
data or collaborating with a team.
Making Free Pay. Subscribers have two key user journey entry
points: 1) a trial 2) free. Streaming services need to make better use
of their analytics (which are increasingly sophisticated) to identify
which free users to invest time and effort into trying to convert. For
example, Spotify or Deezer use free as a marketing tool. So they
have no reason to cling to free users that show no sign of converting.
Instead after a sufficient period of free music has been offered users
should be pushed to subscriptions or onto a radio tier (see figure).
There is no business benefit to the streaming services nor rights
holders to have perpetual on demand free users.
Streaming Music Industry Analysis
This industry is young but growing dramatically. Currently
Pandora and Spotify are the biggest players but the industry
is getting more competitive because of new entrants.
(Statista)
APPENDIX
4. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
Comparison of top 2 music streaming services (from HPs)
Pandora Spotify
Total active users 76.5million (Q3 2014) 75million (Nov 2015)
Premium subscriber fraction About5% About25%
Premium fee $4.99/month $9.99/month
Service Music radio, not on-demand On-demand
Restriction of free users Advertisement between songs
Limited music skip
10hours/month; Occasional Ad
interruption; No off-line listening
Average Revenue $1.0/month/users $3.4/month/users
Financial information (2014) Revenue $92million
Profit -$30million
Revenue $1.1 billion
Profit -$182 million
Spotify is the most successful freemium music streaming business. Spotify's success as a freemium product can be
attributed to the breadth and ingenuity of its product portfolio and the virality derived from social features1
.
- Conversion to Premium. Upon its initial launch, Spotify's product strategy involved engaging with the user
on an unpaid basis, establishing regular use patterns, and then enticing the user to upgrade to a paid account
tier. By bundling cross-device functionality based on payment—paid accounts provide access through an
unlimited number of the user's devices—the most engaged users are prompted to upgrade in order to access
their music whenever they want.
- Retention. And once users had upgraded, they were incentivized to keep their subscriptions current through
Spotify's offline mode, which only allows a user access to their downloaded songs while subscription is active.
- Virility. In terms of distribution, Spotify's viral, social strategy served as the engine of its initial growth. The
ability to link to objects within the application allows users to simultaneously share the efforts of their playlist
creation and incite new users to adopt the product. And through its partnership with Facebook, Spotify allows
its viral dispatches to be associated with the feel-good sentiment of new artist discovery: what are essentially
advertisements for the service on Facebook, can be interpreted by their recipients as a courtesy by Spotify.
Digital Marketing Channels
Digital marketing spending has been growing and is
forecasted to hit $76.6 billion in the US in 2016. Search
and display will continue to have the biggest proportion of
digital spending, comprising 44% and 36%, respectively2
.
In terms of effectiveness of different digital marketing
channels, recent research and surveys have found email
and organic search / search engine optimisation (SEO)
to be the most effective. This is shown in research
compiled by Conductor3
, as indicated by the most popular
digital marketing channels via which people discovered
new websites.
Comparison of Digital Marketing Channels
Digital Marketing Channel How people discover new websites Source of customers for
E-Commerce and B2C Firms
Organic Search / Search
Engine Optimisation
54% 18%
Paid Search / Search Engine
Marketing
18% 10%
Social Media Marketing 32% 1%
Email Marketing 51% 7%
1
Freemium Economics—Leveraging Analytics and User Segmentation to Drive Revenue
2
Forrester Research Interactive Marketing Forecasts, 2011 to 2016 (US)
3
Lam, B. (2014, April 17). [Infographic] Which Marketing Channel Gets You the Most Bang for your Click? Retrieved Nov 21, 2015, from
http://www.conductor.com/
5. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
A study by Gigaom and Extole4
(based on a survey of 300 US-based digital marketers) found email marketing,
social media marketing and SEO to be one of the most popular tactics for achieving brand awareness. Another study
by Ascend25
(based on a survey of 333 marketing, sales and business professionals globally) found email marketing,
social media marketing and SEO to be rated the most effective types of digital marketing.
An interesting finding from the Gigaom/Extole study was that referral marketing (e.g. refer-a-friend programs)
which is being integrated with social and email marketing, enables marketers to get value for acquisition and
conversion, while rewarding existing customers and generating new customers. It is considered extremely effective,
generating a higher percentage of new customers than other tactics.
The studies above, however, do not distinguish between the
types of products/ services being sold. In relation to the
music streaming industry, it may be useful to consider the
effectiveness of digital marketing channels for
marketing mobile apps, since users are likely to use
digital music streaming via apps installed on their mobile
devices. In this regard, an AppsFlyer study6
found that
social app marketing, which includes app marketing on
social networks like Facebook and Twitter, and users
inviting their friends, was the top app marketing
channel when considering three metrics: user quality,
conversion rates and volume. Specific to mobile apps, app
store optimization is also becoming increasingly important.7
Insights8
1) Expand difference between free and paid service. One of the chief purposes of freemium is to attract new
users. If you're generating lots of traffic but few people are paying to upgrade, you may have the opposite
problem: Your free offerings are too rich, and it's time to cut back.
2) Ensure that customers fully understand the premium offer. It could probably monetize more users if the
distinctions between free and paid offerings are clearer.
3) Increase in free users eventually leads to larger paid users. Focus on building a strong, active community.
Freemium’s core growth engine is social proof aka word-of-mouth marketing. All other things being equal, it
4
Nanji, A. (2014, July 15). The Most Effective, Most Used, and Most Budgeted for Digital Marketing Tactics. Retrieved Nov 21, 2015, from
http://www.marketingprofs.com/
5
Marketers Continue to Rate Email the Most Effective Digital Marketing Tactic. (2014, Sep 23). Retrieved Nov 21, 2015, from
http://www.marketingcharts.com/
6
Beese, J. (2013, May 10). Social Media Is More Effective at Marketing Mobile Apps [Study] | Sprout Social. Retrieved Nov 21, 2015, from
http://sproutsocial.com/. According to the study, social media marketing campaigns were especially effective for games, travel, social apps, and
geo-targeted apps, while search delivered the highest quality users for e-commerce, utilities, and food apps.
7
Dogtiev, A. (2015, Jan 29). App Marketing: What Channels Will Be Most Effective in 2015? Retrieved Nov 21, 2015, from
https://appdevelopermagazine.com/
8
Making "Freemium" Work. Harvard Business Review. May 2014, Vol. 92 Issue 5, p27-29. 3p. 1 Color Photograph, 1 Illustration, 1 Chart, 1
Graph.
6. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
would be better to convert 5% of 2 million monthly visitors, for example, than to convert 50% of 100,000
visitors. The best long-term strategy is generally to aim for a moderate conversion rate (research has shown
that most companies' range from 2% to 5%) coupled with a high volume of traffic.
4) Referral & Social App marketing. Referral marketing enables marketers to get value for acquisition and
conversion, while rewarding existing customers and generating new customers. Social app marketing, which is
relevant to music streaming, performs well in terms of user quality, conversion and volume.
Quantitative Analysis
Cleaning Data
To prevent outliers from skewing the results of the analysis, the dataset was reviewed to remove users who were
outliers for each of the independent variables for which the value far exceeded the range of values of the rest of the
users. The following users were removed from the data set:
• davedim and mixounette: 29,371 and 65,872 shouts vs range of 0 to 8,892
• siesahorasi, mikeypleasure and billpa: 1,000,000 songs listened vs range of 0 to 138,400
• catachresistant and keiyo: 12,309 and 15,185 posts vs range of 0 to 8,592
Ratios Analysis
Comparing Adopters and Non-Adopters
Current Period Delta1
Current Period Variable Name Adopter
Mean
Non-
Adopter
Mean
Ratio Adopter
Mean
Non-
Adopter
Mean
Ratio
Demographics age 26.30 24.32 1.08 1.48 0.38 3.91
male 0.72 0.62 1.15 0.20 0.22 0.88
Friends friend_cnt 28.38 11.36 2.50 0.00 0.00 0.40
avg_friend_age 25.85 24.53 1.05 0.21 0.05 3.90
avg_friend_male 0.65 0.64 1.02 -0.01 -0.02 0.79
friend_country_cnt 5.38 2.71 1.99 1519.09 553.81 2.74
subscriber_friend_cnt 1.25 0.29 4.36 12.34 2.82 4.37
Content
Consumption
songsListened 25959.55 12431.48 2.09 1.30 0.02 53.77
Content
Organization
lovedTracks 226.13 68.60 3.30 0.02 0.00 8.77
playlists 1.15 0.46 2.50 5.05 0.38 13.25
Community
Participation
posts 16.72 2.16 7.74 1.48 0.38 3.91
shouts 73.45 16.54 4.44 0.20 0.22 0.88
The ratio analysis shows that the following variables are likely to be associated with conversion to premium
subscription:
- friend_cnt
- friend_country_cnt
- subscriber_friend_cnt
- songsListened
- lovedTracks
- playlists
- posts
- shouts
- delta1_friend_cnt
- delta1_friend_country_cnt
- delta1_subscriber_friend_cnt
- delta1_song_listened
- delta1_lovedTracks
- delta1_posts
- delta1_playlists
- delta1_shouts
From the qualitative analysis, we understand that building a strong, active community is important in improving
viral / word-of-mouth marketing. This implies that the following variables may be important in predicting
conversion to premium subscription: shouts, posts, friend_cnt, songsListened
Therefore the qualitative analysis supports the quantitative analysis in terms of the variables which could be
predictive of conversion to premium subscribers.
Correlation Analysis
The correlation matrix for the variables is shown in the table at the end of the Appendix. Based on the analysis, we
find that the following variables are relatively highly correlated (ρ > 0.5):
7. 15.570: Digital Marketing & Social Media Analytics – Final Project
Group: Shuichi Maeda, Rahul Muchhal, Chung Wei Lee
friend_cnt and friend_country_cnt [ρ = 0.737]
friend_cnt and subscriber_friend_cnt [ρ = 0.762]
friend_country_cnt and subscriber_friend_cnt [ρ = 0.554]
avg_friend_age and age [ρ = 0.676]
delta1_friend_cnt and delta1_friend_country_cnt [ρ = 0.667]
shouts and delta1_shouts [ρ = 0.859]
delta1_posts and posts [ρ = 0.528]
delta2_friend_country_cnt and delta2_friend_cnt [ρ = 0.514]
Regression Model
The data provided includes demographic, social and behavioural data of 107,000 users for the current period as well
as the change in this data between the previous and current period (delta1) and the current and next period (delta2).
To create a prediction model for conversion to premium subscription, we will carry out logistic regression with the
dependent variable being adopter and the independent variables being the current and delta1 variables.
For the purpose of building the regression model to conversion to premium subscription, we should not use
independent variables which are relatively highly correlated. Therefore, combining the ratio, qualitative and
correlation analysis above, we use the following independent
variables for the logistic regression model:
Where independent variables are correlated, we have retained the independent variable for which the absolute
adopter/non-adopter ratio is the highest. On this basis, we have excluded the following independent variables:
friend_cnt, friend_country_cnt, delta1_posts, delta1_shouts, delta1_friend_cnt.
The equation for estimating the probability of converting to a premium subscriber is as follows:
U(Subscribe) = α0 + α1(subscriber_friend_cnt) + α2(songListened) + α3(lovedTracks) + α4(playlists) + α5(posts) + α6(shouts)
+ α7(delta1_subscriber_friend_count_cnt) + α8(delta1_friend_count_cnt) + α9(delta1_song_listened) +
α10(delta1_lovedTracks) + α11(delta1_playlists) + Ɛ
The output from the logistic regression (in Stata) is shown below. For the variables delta1_subscriber_friend_cnt
and shouts the 95% confidence intervals given in the output included 0, i.e. there was no confident at the 95%
confidence level that the coefficients are different from zero. Therefore these variables have been excluded.
Based on the results of the logistic regression, we find that the variables playlists and subscriber_friend_cnt were
most strong associated with conversion to premium users, as shown by the high odds ratios.
- subscriber_friend_cnt
- songsListened
- lovedTracks
- playlists
- posts
- shouts
- delta1_subscriber_friend_cnt
- delta1_friend_country_cnt
- delta1_song_listened
- delta1_lovedTracks
- delta1_playlists