Sales Director EMEA of Adjust Fabian Schaeffer presented his keynote at the international mobile marketing conference TargetSummit Moscow Early 2017 (http://targetsummit.com).
2. ������AD NETWORKS
ADVERTISERS
AGENCIES
DISCOVERY
ENGAGEMENT
CONVERSION
VIEW AD
CLICK AD
DOWNLOAD APP
OPEN APP
ENGAGED
HIGHLY ENGAGED USER
YOUR DATA
“Half of the money I spend
on advertising is wasted;
the trouble is I don’t know
which half”
– John Wanamaker
Navigating the
complexity of a
fragmented
ecosystem
+ Complicated Network integrations, multiple SDKs, reportings, double payment
for same user
3. AD
NETWORK
Fraud PreventionDeliverables Cohorts
This month
Network_B
Tracker
Organic
Network_I
Network_C
Network_F
Network_D
Network_B
Network_E
Network_H
Network_J
Network_G
Network_A
TOTAL
1,800
1,600
1,400
1,200
1,000
800
600
400
200
0
535
Installs Linear
Network_D
764
Network_G
Network_J
558
578
Network_H
1,667
Network_I
837
23.10.2016
01.10 05.10 09.10 13.10 17.10 21.10 25.10
France Australia
PLATFORMS
VIEW BY
COUNTRIES
Select platforms
Tracker
The power of mobile
attribution
Discover your best value users
2017:
268bn Downloads
77bn in app revenue
BUT
FRAUD
4. 1. What are we up against - and how do we fight it?
Who is to blame?
5. The three different types of fraud
Click
Spam
Simulated
Devices
Servers Faking
SDK Traffic/Installs
Some App
*emarketer data
Risk for Mobile Ad fraud
Mobile Ad benchmarks worldwide: Share of attempted ad fraud, by
app vertical, 2016*
6. + Servers pretend to be apps and talk to
analytics platforms. You can fake server calls
by sniffing out the connections your device is
making using free tools like Wireshark.
Fraud
+ Analytics SDK with SSL encryption
+ Shared secret
Prevention
Faking HTTP calls to trigger
false installs
+ Publishers are paid for fake installs
+ Too many installs counted
+ Retention rates very low
Effect
7. Simulated installs & behaviour
+ Exclude all IPs from known data centers, proxies,
Tor exit nodes or cloud providers from attribution.
Fraud
Solution
+ Devices are simulated with full OS stack or are
triggered by "mechanical turks" to create legit install
requests.
Effect
+ Installs (and events) attributed to fraudulent publishers
+ Geo spoofing
+ Undercounted retention rates
8. + Limit number of clicks
considered for
fingerprinting from
single IPs.
+ Detect extremely low
yielding campaigns
and deliver landing
pages with Javascript
to create redirects and
stop crawlers.
Fraud
Solution
Background clicks (Click spamming & Preloading)
+ Apps that, without the users
interaction, crawl ads and click
through any URL they find to spam
fingerprinting and claim organic
traffic.
Effect
+ Organic users are attributed to
publishers
+ Ads show very strong in-app retention
& engagement (organic)
+ Very low conversion rates
9. Purchase Verification
protects your revenue
numbers, secures
accurate reporting,
prevents wrongful CPA
payouts
89% of in-app
purchases are fake
30%
Valid Invalid
Valid Invalid
70%
4% 96%
In-app purchase fraud happens when
USERS hack apps and fool the app into
thinking that they paid for the goods when
they actually didn’t.
10. Before the fact vs.
after the fact!
Exploited ad
budgets are only
one of the problems
caused by fraud
• Organic user activity is poached
• Un-real view of campaign
performance
• No benchmarking possibilities
• Lost budgets
11. I now see that the fraud prevention suite is serving a
different purpose beyond just preventing a ‘bad’ install:
It’s my insurance policy when working with new partners.
We can try new things, and I’m free to experiment because I
trust the data.
We can go big, and not hold back.
An Vu
User Acquisition Lead at Rovio
12. 2. The market is moving beyond just installs
of marketers increased their KPIs with personalisation94%
13. 95% of apps are abandoned after
the first few months **
58% of users will churn in the first
30 days of using an app ***
62% will use an app less than 11
times***
LOST USERS &
DECREASED
ENGAGEMENT
=
biggest fear of APP
developers
14. Measuring your user
behaviour by the hour
in any timezone
• Measure burst campaigns and user
behaviour in greater granularity: i.e.
Friday is a good day but when
exactly?
• True retention and conversion rates
15. ‣ Reach more prospects
‣ Increase relevancy/conversion rates
‣ Create tailored marketing programs
‣ Improve customer relationship
‣ Chance user behaviour
‣ Improve ROI
Segmentin
g to
16. Create highly targeted user segments
on the fly powered by your own data
without sharing all your data
Being in control of
creating your user
segments
17. ‣ Lack of control
‣ Data oversharing
‣ Spamming your users
‣ Human error
Don’t
compromise on