What is the Attribution Period?
Glossary
'Attribution Period' refers to the time frame in which Airbridge's Attribution Model is applied. [Airbridge Attribution Models]
You can configure the attribution period on a Service or Channel level. Airbridge provides two types of windows; Lookback Window for app installation and Attribution Window for in-app events.
Types of Attribution Period
Airbridge's Attribution Period is defined differently based on the time of app installation.
- Before App Installs: Lookback Window is used for Ad Touchpoint (Click/Impression) attribution for app installations.
- After App Installs: Attribution Window is used for in-app events attribution after app installation occurred.
Lookback Window
Lookback window is used to determine which touchpoint (click/view) induced a user to install an app during the configured period. Attribution solutions like Airbridge use the lookback window to determine the ads that triggered the app installs and measure the success of ad campaigns.
For example, it is not reasonable to believe that the app install that occurred today is the direct result of an ad viewed a year ago. But what if it was viewed about a month ago or a week ago? Some may say it is attributable but it is still ambiguous. This ambiguity can be resolved by configuring the lookback window.
The lookback window can be configured differently by Touchpoint type and Matching method respectively.
Touchpoint
The touchpoints for app installation are 'Click' and 'Impression'.
Matching Logic
Airbridge employs two different identity matching methods to identify the users belonging to pre-installation and post-installation.
- Device Matching: This method is approximately 100% accurate. It includes Google-Play Referrer, Query string parameter, deeplink, ADID matching, and platform matching.
- Probabilistic Modeling Matching: It collects non-unique information such as a device model name, OS, or IP address, and then matches the user who performed an event to the user who created the touchpoint.
Matching Category |
Technology Type |
Accuracy |
| Device Matching | Android Referrer | approx. 100% |
| ID Matching (GAID, IDFA) | approx. 100% | |
| Platform Matching | approx. 100% | |
| Deeplink Matching | approx. 100% | |
| Probabilistic Modeling Matching | Probabilistic Matching | approx. ~85% |
Find out more details on Airbridge Identity Matching Logic
Channel Description
Global
Attribution Window applied for all the channels found on Airbridge. If no specific configurations are applied for 'Global', the Airbridge default setting will be applied.
Channel
- Integrated Ad Channel
- Paid-Ad channel. You can select the channel name from the drop-down list.
- Custom Channel
- Custom channels include Owned Media and Earned media such as email, blog, SMS, comments, or brand page.
How to Read Lookback Window: Global
Global: The setting applies to the entire channel being tracked.
- The lookback window chart displayed applies only to the device matching method.
The table can be interpreted as below.
- Click - Device Matching: the lookback window for 'click' by the device matching method is 7 days.
- Click - Probabilistic Modeling Matching: the lookback window for 'click' by the Probabilistic Modeling Matching method is 1 day.
- Impression - Device Matching: the lookback window for 'impression' by the device matching method is 1 day.
- Impression - Probabilistic Modeling Matching: the lookback window for 'impression' by the Probabilistic Modeling Matching method is 6 hours.
How to Read Lookback Window: Per Channel
You can configure the lookback window for each specific channel. The above example shows that 'Click' by Device Matching is set to '4 days' which is shorter than the period of '7 days' of Global. Hence, when Airbridge reviews the attribution for app installation for 'Sample Channel', the click touchpoint that occurred 5 days before the app installation date will be disregarded. The rest of the attribution window settings (Probabilistic Modeling Matching, the impression of matching/Probabilistic Modeling Matching) is set the same as Global.
We don't recommend configuring the lookback window separately from other channels unless it is specifically required to analyze the performance of certain ads. It needs to be carefully modified as the lookback window has a significant direct impact on the performance of ads.
What the Graph Shows
The above image shows that the click device matching is valid whereas other metrics are invalid. The 'click' (in the spot of mouse cursor placed) can be reviewed by the device matching method, but not by Probabilistic Modeling Matching.
Glossary
Valid: The touchpoint can be reviewed for attribution. For instance, if the lookback window is set to 7 days, and there are 10 touchpoints that occurred within that period, they all can be reviewed for attribution. Out of these 10 touchpoints, only 1 touchpoint can be the 'winning touchpoint'.
Invalid: The touchpoint cannot be reviewed for attribution.
Last Changed
- The "Last Changed" column records the last changed date of the channel's attribution window if any. If you click the arrow icon next to the tab name, you can find more details.
- If the 'Last Changed' under Global is displayed as “-”, it means it never has been changed and is thus using Airbridge's default setting.
Attribution Window
Attribution Window is a period of time after an app is installed within which an in-app event can be attributed to the app installation. Depending on the target event, two types of attribution windows exist.
Attribution Window for Install
Let's suppose an app was installed (A) from clicking an ad (B) within the lookback window. In this case, the attribution for install goes to the Channel that generated the click touchpoint. Also, this channel will take the credit for any other in-app events that occurred within the configured attribution window.
Activities that a user performs within the app are referred to as in-app events. These include, but are not limited to: Sign-up, Sign-in, and Add-to-Cart. Airbridge provides standard in-app events for your convenience but you can also create any desired custom in-app events.
The "attribution window for install" of the Sample Channel is set to 10 days. In this case, after the app installation by an ad of 'Sample Channel', 'Sample Channel' will be credited for any in-app events that occur within 10 days, and any in-app events that occur afterward will be recorded as unattributed.
Attribution Window for Deep-Linking
A deep-linking event refers to an event in which the app is already installed and the user clicks an ad to be redirected to a specific page within the app. Thereafter, should a user perform in-app activities such as "purchase" within the attribution window, Airbridge will attribute it to the deep-linking event.
For instance, for the above scenario of viewing the ad > clicking the ad > app opens > viewing the product info > purchasing product (B), the attribution for the 'purchase' event belongs to Channel A that induced the user to click the ad. However, it would be difficult to conclude when exactly the user decides to purchase a product after seeing the ad. It could be day 1, day 5, or day 10. Hence, you could configure the attribution window in order to attribute in-app events to the right ad channel.
From the above image, it shows that Global is the only channel in the list which means the attribution window for deep-linking for the entire channel (Global) is set to 3 days. In-app events that occur during the three-day window are attributed to the ad that led the user to click on the deep link.
What about Install event and Deeplink event overlaps?
In case when a deep-linking event occurs before the attribution window for installs has not yet expired, the deep-linking event will be given higher priority as explained below.
Any in-app events that occur within 1️⃣ will be attributed to the channel that induced the user to install the app.
Any in-app events that occur within 2️⃣ will be attributed to the channel that induced a user to click on the deep link event (ad)
Any in-app events that occur after the attribution window for deep-linking have expired will be attributed back to the same channel as 1️⃣ .
Any in-app events that occur after attribution windows for install are expired, will be considered unattributed.
The rationale behind why deep-linking events are given higher priority is to measure more accurately the performance of re-engagement ads. Because it's more reasonable to say that the deep-linking event that occurred closer to the point of purchase has more impact on users making a purchase than the ad-inducing users to install the app.
Configuring the Lookback Window
'Global' Lookback Window
To edit the lookback window, click "Edit Window".
- Default: Airbridge Default setting. If no modifications are made to the Global window, the default window will be applied to all channels.
- Global: Attribution Window applied for all the channels found on Airbridge. Unless you have a different configuration for any specific channel, all the channels that you are working on will have settings for the Global window.
- You can always reset the Global window to the default setting.
Lookback Window per Channel
Click the "Setting by channel" button at the bottom of the table, then an editable row will be added.
Click "Select Channel" button to expand the drop-down list and you can search or select the channel you'd like to configure. The list of Integrated Channels is followed by the list of Custom Channels.
Editing Window Interval
Window interval is available in min/hour/day from a minimum of 5 mins to a maximum of 30 days.
- Please be noted that any newly added channel with the same window setting as Global will not be applied separately from the Global settings.
Deleting a Specific Channel's Settings.
To delete the settings for a specific channel, you can click the recycle bin icon on the far right. Please be noted that deleting the channel settings means that the channel window setting will be applied back to Global.
- You must click 'save' button after deleting the channel window in order for the changes to be reflected in the system.
