Sei sulla pagina 1di 6

mobile app tracking

hasoffers

Tracking Methods
for Mobile Applications

Prepared by the
Mobile App Tracking Engineering Team
2012 HasOers, Inc. | All rights reserved

Tracking Methods for


Mobile Applications
Page 2

mobile app tracking

Tracking methods for mobile applications


Table of Contents
1. Introduction
2. Device Fingerprinting
3. MAC Address
4. ODIN
5. OpenUDID
6. Cookie Tracking
7. UDID (Depricated)
8. TRUSTed Mobile Ads
9. Apple IFA and IFV

hasoffers

Tracking Methods for


Mobile Applications
Page 3

mobile app tracking


hasoffers

1. Introduction
Mobile app conversion tracking presents unique challenges not found in traditional desktop
browser advertising. Standard conversion tracking using pixels or postback is cut short when
users are directed to a marketplace, like the App Store, where tracking cannot continue.
Cookie-based tracking with pixels also fails when a user is directed from a mobile browser to
a marketplace for mobile app installs. The mobile app also cannot read cookies set in the
mobile browser without a fairly disruptive user experience.
Standard cookieless tracking methods, utilizing postback could be an ideal solution.
However, passing a transaction ID through an app marketplace to a completed app install
doesnt work for the iOS App Store. When a user is directed to the App Store to download the
mobile app, it is impossible to pass a transaction ID to the marketplace and onto the mobile
app to verify installation. Because an iPhone or iPad app can't receive a tracking token
generated by a click event in the mobile browser, the mobile app can't load a postback URL
to record the conversion. Android apps are somewhat dierent in that Google Play allows a
transaction ID to pass through the install_referrer parameter, which means Android doesnt
have the same tracking limitation.
A number of device identication technologies and tracking methodologies are available to
connect mobile web and mobile app advertising with mobile app installs.
MobileAppTracking uses a proprietary device ngerprinting technology to allow advertisers
to accurately track installs and conversions. There are also a number of device tracking
methods available with MobileAppTracking for matching a series of events to the same
device.

Tracking Methods for


Mobile Applications
Page 4

mobile app tracking


hasoffers

2. Device Fingerprinting
Fingerprinting technologies anonymously match a combination of attributes to a device to
arrive at a high statistical probability that two events with a similar ngerprint are from the
same device. For example, if someone from a specic device prole taps an ad and then a
similar device prole registers and app install 90 seconds later, the probability those two
events were the same device is extremely high.
A device ngerprint is generated by combining several header data points sent by the device.
When a user clicks on an advertisement, a request is sent to the tracking server. The server
gathers the header data points from the device and generates a unique ID for the device
based on the combined data points. This unique ID is generated by the server, anonymous,
and not associated with Apple's UDID or the ANDROID_ID. When an install event happens,
the app noties the tracking server, which generates a ngerprint based on header data
points. If the ngerprint generated at install matches a ngerprint previously logged from
any other event (such as an ad click) the server determines a match and logs a conversion.
The whole process happens in about 300 milliseconds and should be transparent to the
user. MobileAppTracking device ngerprints are currently 94% unique.
3. MAC Address
Media Access Control (MAC) addresses are unique to a hardware device and have become a
common replacement for the recently deprecated UDID on iOS hardware. In some cases,
MAC address-based tracking solutions encrypt or hash the MAC address in an attempt to
make it anonymous. One key benet of using MAC addresses for tracking is the similarity to
UDID tracking methodologies. A MAC address can be identied across applications and
mobile web, allowing for accurate tracking of install events. MAC address tracking also
carries similar privacy concerns to UDID, which may cause Apple to change the conditions
for access to MAC address usage and alter its availability for tracking purposes.

Tracking Methods for


Mobile Applications
Page 5

mobile app tracking


hasoffers

4. ODIN
Open Device Identier Number (ODIN) is a tracking solution developed by the ODIN Working
Group. ODIN is designed for mobile app developers to uniquely identify user devices in an
interoperable manner. ODIN is anonymous in that it is unique to the device without
revealing data about the originating device make, model, or other data. ODIN provides
standarization with an identier that could be used across iOS, Android, and Windows
Mobile. ODIN is generated slightly dierntly based on which platform it refers to. For iOS,
ODIN uses the 802.11N MAC address as the identier seed, Android uses the ANDROID_ID,
while Windows Phone uses DeviceUniqueID. The identier seed is passed through a SHA-1
hash function, with the resulting 40 character string being used as the ODIN-1 value. This
identier can then be integrated into app tracking SDKs.
5. OpenUDID
OpenUDID is a SHA-1 hash value calculated in a manner similar to Apple's own UDID value.
The goal of OpenUDID is to provide a drop-in replacement for UDID which remains
persistent on the device, is accessible and unique accross apps, and is backward and forward
compatible with existing UDID. OpenUDID currently utilizes the PasteBin on iOS devices to
allow for persistent storage across apps. One key dierence between OpenUDID and UDID is
that users can opt out of device tracking via OpenUDID. If OpenUDID is integrated into a
tracking SDK, ad click events and app installs can both be logged with an associated
OpenUDID to allow for accurate tracking of unique devices.
6. Cookie Tracking
Cookie tracking sets a rst party cookie in Mobile Safari on iOS. Cookies on a mobile device
are the equivalent of their counterparts in desktop browswers, though they are more limited
by default mobile device security settings. A cookie results in accurate attribution of an ad
click to the device. Due to the security settings in Mobile Safari, cookies cannot be shared
across apps or shared between a mobile app and the mobile web. The workaround to pair
an app install to an ad click is a brief redirect to Mobile Safari on application launch, which
allows the cookie to be read. Cookie data is lost anytime either the brower cache or HTML5
local cache is cleared. There's also a possibility of duplicate attribution with cookies in a
scenario where the user clicked two dierent ads prior to app install.

Tracking Methods for


Mobile Applications
Page 6

mobile app tracking


hasoffers

7. UDID (Depricated)
Apple's Unique Device Identier (UDID) was the standard mechanism for tracking installs
across mobile web and mobile apps. Prior to being depricated, UDID was included as a
standard call within the iOS SDK. Due to pressure around privacy concerns, Apple
depreciated access to UDID in applications, resulting in apps using UDID being rejected from
the App Store.
8. TRUSTed Mobile Ads
In April 2012, TRUSTe announced their TRUSTed Mobile Ads initiative, which is designed to
aord users the ability to opt out of tracking. While it's not a tracking mechanism itself,
TRUSTed Mobile Ads is an important consideration in the mobile tracking ecosystem.
Tracking SDKs which integrate TRUSTed Mobile Ads with applications will allow users control
over how much of their data can be tracked.

9. Apple IFA and IFV


With the release of iOS 6, Apple rolled out two new identiers for advertising. The
Advertising Identier (IFA) is a non-permanent, non-personal, device identier apps can use
for tracking. The Identer for Vendor (IFV) is an identier used to associate apps that come
from the same vendor on the same device. A dierent IFV value is returned for apps on the
same device that originate from dierent vendors, and for apps on dierent devices
regardless of vendor.

Potrebbero piacerti anche