Glossary
Find definitions of the user IDs, predictions, and API response terms used in CUE.
Context API
The Context API accepts an app user ID and returns predictions and information about the ad credited with that user's app installation. Call the API from your server or CRM and use the results to define when your service takes an action.
API request and response
An API request is a call from your system to CUE that sends a user ID to retrieve information. A response contains the result and data CUE returns. For example, your server sends an app user ID, and CUE responds with predictions for that user.
JSON
JSON is a format for exchanging data through APIs. It pairs names with values. For example, "probability": 0.723 means the field named probability has the value 0.723.
User ID and identifier mapping
A user ID identifies the app user whose information you want to retrieve. Include exactly one of these three fields in each Context API request:
| Request field | ID to send |
|---|---|
userId | The Airbridge Generated UUID that Airbridge assigned to the app user. |
brazeDeviceId | The Device ID from the Braze SDK. |
airbridgeExternalUserId | The external user ID your service sent to Airbridge. |
Identifier mapping connects a Braze Device ID or external user ID to the corresponding Airbridge user. brazeDeviceId is different from Braze's external_id.
App SDK token and web SDK token
An app SDK token or web SDK token authenticates Context API requests. Use a token from the app connected to CUE, and store it in secure server or CRM configuration. Find the tokens under [Settings]>[Tokens] in the Airbridge dashboard.
Prediction probability
Prediction probability (probability) expresses a model's estimate of an event's likelihood as a number between 0 and 1. A value of 0.723 means the model predicts a 72.3% chance of the event. To interpret the probability, check which event and time window the model predicts.
Prediction tier
Prediction tier (tier) groups results according to the distribution of a model's predicted probabilities. CUE returns low, medium, or high. A purchase tier of high indicates a high predicted likelihood of purchase. A churn tier of high indicates a high predicted likelihood of churn. See Using Predictions for details.
Prediction window
A prediction window is the period covered by a model's prediction. The d3_purchase model predicts the likelihood of a standard order-completed event within three days after app installation. The d3_churn model predicts the likelihood of no activity events after day three through day 30 after app installation. Both models use the app installation time as the reference point.
Installation attribution
Installation attribution identifies the advertising touchpoint credited with an app installation event. CUE returns its channel, campaign, ad group, ad creative, and configured ad keyword in context.lastTouchpoint. This object describes the touchpoint credited with installation, rather than the most recent ad interaction or every ad the app user has seen. See Using Installation Attribution to connect this information to the ads you run.
null
In JSON, null represents a missing value. It is different from prediction probability 0 or tier low.
When no installation attribution result is available, the entire context.lastTouchpoint object is null. When only an individual value is missing, that field is null. The values unattributed and restricted are different from null. The value unattributed alone does not establish organic acquisition.
Default behavior
Default behavior is the action you define for your service when it cannot use an API result. For example, your service might show existing guidance or use a default message. Implement this behavior for API errors and timeouts as well.
HTTP status code and requestId
An HTTP status code indicates the result of an API request. The Context API returns 200 when it processes a request successfully. For errors, check both the status code and the response body.
The requestId identifies an individual API request. Use it when investigating errors or contacting Airbridge. See Error Handling for details.