Attribution Models

Users are exposed to numerous advertisements prior to installing apps or making in-app purchases (conversions). In this user journey, we call advertisement clicks or view activities 'Touchpoints' and a model determining which touchpoint attributed to a specific conversion an 'Attribution Model". 


Airbridge's Attribution Model

Different types of attribution models include Last Click, First Click, Linear and etc. Airbridge uses Last Click model as a base model and added attribution qualifications and attribution priorities as additional features. We also have created multiple rules for different types of conversion events (installation, deep-link, app opening/ in-app events). Even if a touchpoint meets all of Last Click requirements to be accredited, we do not consider it an attributed touchpoint unless it satisfies requirements for attribution qualification or if another touchpoint has higher priority. We call this accredited touchpoint a 'Winning Touchpoint'. 


1. Attribution Qualification

  1. Amongst countless touchpoints, only the ones that fall under lookback windows (or attribution windows) are subject to being selected as a Winning Touchpoint.


    • Lookback windows are a set of time frame that leads up to a specific conversion event (installation, purchase, and etc) 
    • Lookback windows determine the pool of touchpoints we review to select a winning touchpoint. For instance, if we decide to use a 7days lookback window, all touchpoints that occur within seven days prior to the conversion will be reviewed for a winning touchpoint.
    • Number of days in the lookback window can be changed in Airbridge dashboard
      • Find out more about Airbridge lookback window → Link (TBD: docs)
  2. Fraudulent touchpoints will be dismissed from user journeys.

    • Airbridge provides various features to analyze and eliminate fraudulent touchpoints.
    • Once the fraud touchpoints criteria are set up, you may later view the corresponding fraud touchpoints that meet the criteria.
      • Airbridge fraud touchpoints prevention features → Link (TBD: docs)


2. Attribution Rules for Installations

We prioritize attribution for the installation based on the rules below. Should multiple touchpoints hold equal priority, we then select the winning touchpoint based on the Last Click model (The most recent/last touchpoint among equally prioritized touchpoints will become the winning touchpoint). 

  1. Touchpoints are prioritized based on their channels

    1. Airbridge gives first priority to touchpoints that occur in Self-Attributing Network (Google Ads, Apple Searchads, Facebook and etc) or tracking links. 
    2. The next (second) priority goes to touchpoints that are collected by UTM Parsing. 
    3. Third priority will be to CTA (call-to-action) feature in SDK-installed web (if setDownloads and setDeeplinks have been used) → Link (TBD)
  2. Clicks always win higher priority over views. 

  3. Deterministic Methods (Referrer, Cookie Matching, ID Matching, and etc) always win higher priority over Probabilistic Methods (fingerprinting) as deterministic methods have a higher accuracy rate. 

    • Airbridge User Matching Method → Link (TBD)


Attribution Scenario for Installation


  1. We only consider touchpoints that fall under lookback window → Eliminate 'A 채널 클릭' 
  2. We give higher priority to clicks than views → Eliminate 'B 채널 조회'
  3. We use Last Click model if the touchpoints are equally prioritized →Eliminate 'B 채널 클릭'

⇒ Winning touchpoint is 'C 채널 클릭'


3. Attribution Rules for Deeplinks


Deeplinks are most commonly generated after a user clicks on a tracking link. Therefore, a click that incurs this deeplink becomes a winning touchpoint. 


4. Attribution Rules for In-app Events or App Openings

Airbridge defines in-app events or app openings as events that occur only after when deeplink is generated. Hence, we accredit any installations or deeplinks that occur immediately before the in-app or app opening events. 

Attribution Scenario for In-app or App Opening Events

  • Scenario 1


    • Above is the most common attribution scenario for in-app events. As displayed, sign-up and log-in touchpoints both fall under the lookback window of 30days, allowing both touchpoints to be the attributing touchpoints. On the other hand, we eliminate CustomEvent1 and Add-to-Cart as they are stationed outside the lookback window. 
  • Scenario 2


    • Lookback windows (attribution windows) for in-app events are set for 30 days by default while deeplink lookback windows are set for three days. You can see above that log-in and purchase events both fall under the 30day lookback window, allowing them to be the attributed touchpoints.
    • Let's try to find a winning touchpoint of add-to-cart event in the user journey above. We see that the deeplink event occurred immediately before add-to-cart event (also falls under the three day window). This tells us that the add-to-cart's winning touchpoint is in fact not the installation event but a deeplink event. 
Was this article helpful?
0 out of 0 found this helpful
Current Page