The methods performed in MyTracker work the same way for device and user attribution. The only difference is the attributable event: for devices — the installation or the first site visit, for users — the first event passed with the user ID.
MyTracker supports all major mobile attribution methods: Android Referrer, Device ID matching, Probabilistic.
Android Referrer works only with Android apps distributed through Google Play. To use this attribution method, the link to the app store must contain a special parameter, referrer, whose data identify the source of referral. Once an app is installed, it will receive the required parameter value through Android OS services. Then, that value will be transmitted to a tracking system server and used to detect the source of installation.
Android Referrer is the most preferred attribution method because of its outstanding accuracy. Android Referrer data are used for attribution, even if there are other clicks closer to the time of installation.
To attribute installs from Facebook, you need to pass an Install referrer decryption key to MyTracker. Learn more
This method is used when the platform where a tracking link is placed is capable of obtaining user's device ID and transmitting it to the tracking link. The following identifiers are most commonly used for this purpose:
When a user clicks a tracking link, their device ID is captured and saved by the tracking server. After your app has been installed, the tracking library within the app receives that ID value and transmits it to the tracking server, where it's used to find a click record with a matching identifier.
Device ID Matching is a fairly accurate and the most common attribution method. It works very well for in-app ads but is not usable when the tracking link is placed in an environment that is not capable of obtaining device identifiers. For example, email and SMS campaigns, print ads, web pages accessed through a browser, etc.
Probabilistic attribution uses machine learning models. This method identifies the installation source without using advertising device identifiers (IDFA, GAID, etc.), maintaining user privacy.
Probabilistic modeling, based on a large amount of historical data, continually learns from newly available stats from devices that have provided access to the ad identifier (in particular, through App Tracking Transparency on iOS) and SKAdNetwork stats.
The probabilistic model assumes that an install relates to a particular ad campaign and partner based on device characteristics and information about advertising interactions. Because this method uses probability estimation and multiple device parameters, the attribution time for installs can take longer (about 1 hour on average).
MyTracker allows you to restrict or to disable Probabilistic, when, for example, you know that other attribution methods are in use.
Since 2021 Fingerprint is disabled due to Apple's privacy updates
Fingerprint operates pretty much like Device ID Matching: when a user clicks on an ad and is redirected, the tracking server receives some user data (for example, the user's IP) that are available at that time. After your app is installed, the tracker SDK collects the same set of data and passes it to the server to search for a matching click event.
The difference is that, as a rule, the available set of data is not sufficient for pinpointing the actual end user. For example, several users associated with the same IP may follow an ad link and install the app.
Due to its constraints, this method is less accurate than others. MyTracker allows you to restrict or to disable Fingerprint, when, for example, you know that other attribution methods are in use.
Because of its low accuracy, Fingerprint is the least preferred solution. Fingerprint attribution is used only when no other click data are available.
For web attribution, MyTracker uses two methods: Redirect
and Direct link with the
During redirection in case of data loss,
an auxiliary method of attribution by cookies may work.
All methods allow tracking post click conversion and the only attribution with cookies supports conversion by post view.
The method is based on the transfer of a unique identifier when redirecting to the website:
mt_click_id, adds it to the website URL and also keeps the click data (click time, cookies with user data, etc.).
When a new user visits your website, MyTracker compares the cookies from the site with the same tracking clicks and ad impressions. If the values match, the tracker will associate the appearance of a new user to the ad by the last click/show closest to the view.
If your partner does not support redirect, you can use direct links.
The essence is that MyTracker adds a special parameter
to the target URL. This parameter contains the tracking links identifier.
MyTracker gets the value of this parameter together with the hit (site visit) and links the visitor to the relevant tracking link and campaign.
We recommend using the Redirect as it's the most accurate method. Attribution by Direct links suggests some disadvantages:
mt_link_idin the bookmark, and in case of a long absence (after the inactivity window has expired), when switching from the bookmarks, the reattribution will trigger, and MyTracker will link it to the same campaign.
DOOH (Digital Out-of-Home) attribution is valid only for Android. To link ad impressions and app installs, MyTracker uses MAC address as a unique identifier assigned to a network interface controller.
When an ad appears on digital billboards, our server receives data on all devices in front of each screens. MyTracker gathers MAC addresses, advertising time, and specific parameters embedded in the banner body (these are attribution window, app and tracking link ids). If some user install an app during the dooh window (by default, 1 day) on a device with a tracked MAC address, MyTracker could associate user appearance with out-of-home advertising.
DOOH attribution priority is lower than Device ID matching (post view), but higher than Fingerprint (post view).
For more details about DOOH ads, refer to DOOH tracking