Learn how to:
- track
- compute
- and leverage
Customer Lifetime Value for the user acquisition campaigns of your mobile games.
Featuring industry experts Eric Seufert from Wooga, James Peng from Storm8 and Simon Kendall from adjust by adeven. Get the right users!
2. AppLift.com 2Leveraging Customer Lifetime Value for user acquisition campaigns
About AppLift
300+ Game
Publishers
The leading mobile games marketing platform
400+ Games
2,000+ Media
Partners
We are a team of 85+ mobile game
enthusiasts
We are headquartered in Berlin with
offices in San Francisco and Seoul
We come from 25+ countries and speak
20+ languages
We are backed by Prime Ventures and
HitFox Group with USD 20M investment
100%
mobile
games
LTV focus
All key
markets
3. AppLift.com 3Leveraging Customer Lifetime Value for user acquisition campaigns
Our panelists
Eric Seufert
Head of Marketing, Wooga
Eric Seufert is a quantitative marketer with a passion for blending
real-world problems with large amounts of data, econometric
frameworks, and analytical systems. His professional specialty lies
in programmatic statistical methods and predictive forecasting for
freemium products.
Eric received an undergraduate degree in Finance from the
University of Texas at Austin and an MA in Economics from
University College London, where he was an Erasmus Mundus
scholar. Eric joined Skype immediately out of graduate school and
subsequently held marketing and strategy roles at Digital Chocolate
and Wooga, where he is now the Head of Marketing.
Eric is also the author of Freemium Economics, published by
Elsevier in 2014.
Originally from Texas, Eric currently lives in Berlin. In his spare
time, Eric enjoys traveling and writing.
4. AppLift.com 4Leveraging Customer Lifetime Value for user acquisition campaigns
Our panelists
James Peng
Head of User Acquisition and Monetization, Storm8
James Peng leads global user acquisition and monetization for
Storm8's complete portfolio of 40+ titles. Storm8 is a leading mobile
social gaming network and developer with more than 50 million
monthly active users and 600 million downloads across 300 million
devices worldwide.
Prior to joining Storm8,James worked at H.I.G. Capital, a $15 billion
private equity fund, where he evaluated, structured, and executed
leveraged buyouts in partnership with management teams. Prior to
that, he worked as an investment banker for Lazard, where he
advised on various M&A and restructuring deals.
5. AppLift.com 5Leveraging Customer Lifetime Value for user acquisition campaigns
Our panelists
Simon Kendall
Product Manager, adjust by adeven
Simon is adjustâs Product Manager and all rounder technology
translator.
He assists the CTO, and the sales and technology teams in
transforming product feedback and business analysis into product
engineering. Simon holds regular technology âwhiteboardâ sessions
to support the sales and account management teams with their
product knowledge and trouble shoots complex product requests
from clients. Over the past year he has also managed the technical
integration of adjust with 200 partners and networks.
Simon is bilingual in Swedish and English. He has previously
presented at the 2013 Apps Promotions Summit in Berlin on cohort
analysis and discussed the anatomy of a metric at the 2014 Winter
Nights conference in St Petersburg.
6. AppLift.com 6Leveraging Customer Lifetime Value for user acquisition campaigns
Acquiring the right users
1
2
3
4
What is LTV
How to track and measure
LTV in practice
How to leverage LTV for
user acquisition campaigns
Questions
7. AppLift.com 7AppLift.com
1. What is LTV?
What is Customer Lifetime Value?
How is LTV defined theoretically? What
are its components?
Why is it important for game publishers?
8. AppLift.com 8AppLift.com
1.1 What is Customer Lifetime Value?
Eric:
Conceptually, LTV is discounted expected revenue
contributions from a user. Practically, LTV is the price you are
willing to pay in order to bring a user into your game, given a
set of characteristics of that user.
James:
LTV is the value that a user brings in, including monetary value
(in-app purchases) and additional organic value.
Simon:
LTV is the total value that a user brings during the period from
initial conversion until he/she no longer engages with the app.
9. AppLift.com 9AppLift.com
1.2 How is LTV defined theoretically?
2 conceptual approaches to evaluate LTV:
The retention approach â bottom up
The monetization approach â top down
10. AppLift.com 10Leveraging Customer Lifetime Value for user acquisition campaigns
What is LTV? The retention approach (bottom up)
Source: Profitably launching Jelly Splash to #1, a marketing postmortem. GDC 2014, Eric Seufert
11. AppLift.com 11Leveraging Customer Lifetime Value for user acquisition campaigns
The retention approach (bottom up)
Source: Profitably launching Jelly Splash to #1, a marketing postmortem. GDC 2014, Eric Seufert
Eric:
1. Use a retention profile to calculate customersâ lifetime
2. Use ARPDAU as a monetization component of LTV
3. Multiple the two elements
12. AppLift.com 12Leveraging Customer Lifetime Value for user acquisition campaigns
The retention approach (bottom up)
Source: Profitably launching Jelly Splash to #1, a marketing postmortem. GDC 2014, Eric Seufert
Eric:
This approach is generally used in a prototyping stage of the
game and it is not meant to be very accurate. It gives a rough
estimation of LTV. If you have a high LTV you can generally
spend more to acquire users and if you have a low LTV â
wellâŚyou canât.
13. AppLift.com 13Leveraging Customer Lifetime Value for user acquisition campaigns
What is LTV? The monetization approach (top
down)
Source: Freemium Economics, Eric Seufert
14. AppLift.com 14Leveraging Customer Lifetime Value for user acquisition campaigns
The monetization approach (top down)
Source: Freemium Economics, Eric Seufert
Eric:
This is a more practical model compared to the bottom up approach.
1. Break down groups of users that you are targeting by
various sets of characteristics (e.g., geography, platform,
device).
2. Gather minimal level of historical data about monetization
from that specific group.
3. Project it out based on a logarithmic term.
This model enables to make a conclusion on how much you are
willing to spend to recoup money over a certain amount of time. We
generally use 180 days for casual games.
15. AppLift.com 15AppLift.com
1.3 Why is LTV important for game
publishers?
James:
LTV helps games publishers understand the ability to pay a
different cost per user from every different channel.
The more accurate you are across every single campaign, the
more efficiently you can buy and maximize your targeted
returns with your targeted volumes.
The more precise you are
ď the more powerful in the marketplace you can be
ď the more efficiently you can command volumes from every
different channel
16. AppLift.com 16AppLift.com
2. How to track and measure
LTV in practice
How to measure it? How is it computed in practice?
How to track post-install events? How much is it to set it up?
Which events to track? How relevant are they?
Whatâs the role of cohort analysis in tracking LTV?
How to perform cohort segmentation?
17. AppLift.com 17AppLift.com
2.1 How is LTV computed in practice?
James:
First, you need to have the infrastructure in place to be
able to match the different sources of revenue with the
devices that they belong to.
In practice, I typically track cohort revenues using KPIs (e.g.,
day 1, day 3, week 1, week 2, week 3, month 1).
To calculate the LTV, I take all the history to-date and
map a revenue curve for every game.
18. AppLift.com 18AppLift.com
2.1 How is LTV computed in practice?
James:
Generally LTV curve is similar across most similar
sources. If you have a very strong understanding of how
a user behaves across one particular game, you can
map that across similar channels that you have.
If you do not have any information available, you might need
to use a proxy, such as a title in the same genre, and apply
it on a revenue curve. Otherwise, you will have to use
multipliers and use a pre-build estimate.
19. AppLift.com 19AppLift.com
2.1 How is LTV computed in practice?
Eric:
The biggest mistake that prevents people from getting
insights from their data â not splitting up users by
cohorts.
Looking at users as at a collection of characteristics (the
network they are acquired from, the location, the date they
were acquired on, etc.) helps to group users into groups that
make sense.
You should track the cohorts every time. If you donât, you fall
into a trap where you are averaging users from different age
groups and comparing information that is not comparable
(e.g., users from Basil vs. Norway vs. U.S.).
20. AppLift.com 20AppLift.com
2.2 Tracking post-install events
Simon:
If you have a good tracking solution (in-house or 3rd
party), technical implementation of post-install events
tracking should not be hard.
The most important and the most time consuming
consideration when setting up post-install events tracking
should be: What I am trying to capture? What goals do I
have? How do I capture the data that is most relevant?
While in-app purchases and monetization metrics matter a
lot, other metrics (virality and engagement) are very
important too and greatly vary from publisher to publisher.
22. AppLift.com 22AppLift.com
2.2 Tracking post-install events
Simon:
Events on the previous slide are good guidelines, but
are not necessarily applicable to all use cases.
Tutorial completion is a very interesting event. A lot of clients
are tracking this event. Lots of clients are also going further
than that â specifying level 1 completion, level 3 completion.
When tracking a lot of different KPIs, itâs crucial to
remember what is the central goal in tracking, what is the
core metric and parameter metric.
23. AppLift.com 23AppLift.com
2.2 Tracking post-install events
Simon:
While parameter metrics are nice to have in the
background, ultimately there should be a couple of core
metrics.
E.g., if you need to grow your market share â focus on
retention; if you need to get your money back ASAP â
focus on the purchases.
24. AppLift.com 24AppLift.com
2.3 Whatâs the role of cohort analysis
in tracking LTV?
Simon:
Cohort analysis takes every single metric that you might
want to calculate and gives you a quicker, clearer
feedback.
E. g., an install date cohort: group the incoming users by the
day on which they install. This gives an immediate clear
feedback on how users are behaving right after the install
compared to how previous cohorts might have behaved.
25. AppLift.com 25Leveraging Customer Lifetime Value for user acquisition campaigns
Cohort analysis
Erase interference and track comparable KPIs
Source: Beyond the Install, APS Berlin 2013, Simon Kendall
26. AppLift.com 26AppLift.com
2.3 Whatâs the role of cohort analysis
in tracking LTV?
Eric:
If you target your market, you donât throw your budget on an
ad network without at least some parameters attached to it.
If you are not collecting data and you are not segmenting
around various characteristics, there is no way that you can
do marketing.
27. AppLift.com 27AppLift.com
2.3 Whatâs the role of cohort analysis
in tracking LTV?
Eric:
The obvious examples of cohorts/segments are location and
device: bid a different amount for a U.S. iPhone user and for
a German Samsung Android user.
More specific examples of cohorts: phone model, network
from which users were acquired, the type of ad, specific
events (holidays) during which users were acquired.
28. AppLift.com 28AppLift.com
3. How to leverage LTV for user
acquisition campaigns?
Why should game publishers leverage their LTV data?
Which LTV approach is best for user acquisition purposes?
What hinders game publishers from using LTV for campaign
optimization?
Does it require a lot of internal resources?
How do the game genre and features come into play?
29. AppLift.com 29AppLift.com
3.2 Which LTV approach is best for
user acquisition purposes?
There are 2 general approaches for LTV:
1. Predicting whatâs the LTV, the number in dollars of the
users that you acquire.
2. Looking at post install events (proxies) and compare
which ones perform better than others.
30. AppLift.com 30AppLift.com
3.2 Which LTV approach is best for
user acquisition purposes?
James:
In practice you need to use both approaches, but it also
depends on how much data and time you have.
When you are starting fresh, you need to use predictive
analytics to forecast without any data at hand.
If you are fortunate to have other title that you are marketing
and have multiple data behind it, you can quite accurately
estimate usersâ value curve. Thus, you can leverage the
data you have and insights you have from other titles.
31. AppLift.com 31AppLift.com
3.2 Which LTV approach is best for
user acquisition purposes?
Eric:
If you set up a model to forecast based on a minimum level
of data (2-3 days), you need to continuously update this
model as you accumulate more data.
LTV prediction made up on 3 days data will change on the
day 5 or 6, so you need to be able to accommodate new
data sets.
32. AppLift.com 32AppLift.com
3.2 Which LTV approach is best for
user acquisition purposes?
Eric:
Getting back to âbottom upâ approach â there is no way to
forecast LTV based on some minimum set of data. Even if it
has been an internal launch with 30 people, it is not valid
data.
I like to have at least a thousand data points before doing
any modelling.
33. AppLift.com 33AppLift.com
3.2 Which LTV approach is best for
user acquisition purposes?
Eric:
You have to also weight the uncertainty against the money
you are getting. If you are not doing aggressive marketing
because you are building up a dataset, then you are not
acquiring users.
There is some level of uncertainty you have to be
comfortable with, in order to be aggressive enough in
marketing a game that you believe in.
34. AppLift.com 34AppLift.com
3.3 What hinders game publishers
from using LTV for campaign
optimization?
James:
Itâs difficult to understand the true LTV: game changes, the
channels that you are buying keep changing as well (they
are varying composition, keep swapping up publishers in
and out).
Even more difficult part is engineering behind it. It can be as
hard as a home grown solution or as straightforward as a 3rd
party solution (e.g., game analytics), but definitely
engineering is the biggest problem.
35. AppLift.com 35AppLift.com
3.4 How do the game genre and
features come into play?
Eric:
The game genre doesnât fundamentally change the concept
of LTV or how it is used in marketing.
However, the genre and the appeal of the genre and total
addressable market affect aggregate levels of profit and
your ability in the market.
36. AppLift.com 36AppLift.com
3.4 How do the game genre and
features come into play?
Eric:
That can be teased out in a soft launch, but if you wait too
long to make a determination, then you are not being
strategic. Thatâs why LTV modelling without any data comes
into play.
You need to make corporate level decisions about the game
in terms of how much you will market it, given a total
estimated addressable market and the opportunity cots of
building that game.
37. AppLift.com 37AppLift.com
3.4 How do the game genre and
features come into play?
Eric:
Marketing works from very early stages of prototyping. You
need to estimate early on how big the market could be, how
this game will monetize, what size opportunities you can
achieve.
Genre doesnât necessarily drive the LTV, but your
âguestimationâ of LTV drives how you should approach the
development.
39. AppLift.com 39AppLift.com
Is it possible to track customized
events or should you track standard
events?
Simon:
With a good infrastructure itâs always possible to track
whatever you want.
There is a standard set of events that we see very
commonly, that would be normal solutions (cited in slide 21).
You do have to look at your specific model, your specific
business, goals and then in such a way all events become
custom.
40. AppLift.com 40AppLift.com
How does the fact that you do
marketing and user acquisition affect
LTV within your game? (1)
Eric:
Users are finite resources that you are trying to acquire to
your game. As you have acquired more, the remaining ones
become more expensive because they are more scarce.
The user that you have to be the most aggressive about
acquiring will be less interested in your game, monetize
worse, provide less vitality. You canât compare the LTV of
one cohort where you are running low volume campaigns
against another cohort where you are running high volume
campaigns. These are two separate, not comparable things.
41. AppLift.com 41AppLift.com
How does the fact that you do
marketing and user acquisition affect
LTV within your game? (2)
James:
If you are buying a lot of installs, this can affect your
average user value.
If you are doing a lot of marketing and increasing your
visibility, then you can get exposure to different segments of
users. You can get on top of rankings and get a general pool
of users that find the game because itâs on the chart.
42. AppLift.com 42AppLift.com
What are the biggest challenges in
future developments lying ahead of
LTV and how do you plan on
addressing them? (1)
James:
Predicting and adapting LTV is a continuous struggle. As
developers and publishers we need to continue changing
the definition of LTV â products change and market changes
(internal and external factors).
43. AppLift.com 43AppLift.com
What are the biggest challenges in
future developments lying ahead of
LTV and how do you plan on
addressing them? (2)
Eric:
LTV has been in academic research for 30 years.
Conceptually itâs not changing, itâs fundamental.
But as games become more mainstream, trackability could
be a challenge. It will be challenging to put a price on the
individual download which was inspired by a billboard.
44. AppLift.com 44AppLift.com Berlin | Seoul | San Francisco
Contact
Thomas Sommer
Content Marketing
tso@applift.com
Eric Seufert
Head of Marketing
eric.seufert@wooga.net
James Peng
Head of
UA & Monetization
jpeng@storm8.com
Simon Kendall
Product Manager
simon@adjust.com
45. AppLift.com 45Leveraging Customer Lifetime Value for user acquisition campaigns
Thanks for attending!
Download this presentation, our
case study⌠on:
www.applift.com/INSIGHTS.html
For more tips on global user acquisition,
global expansion etc. get in touch!
info@applift.com