Receiving Scoring Updates Through Callbacks

You are able to automatically receive updated user scores using HTTP requests (callbacks) that are sent right after the score is updated.

Callback data format

Service sends POST requests with JSON body. Example:

{
    "datetime": "2025-07-03T13:13:53.533188", // Callback time
    "data": [
        {
            "user": "p-478823",
            "model_name": "churn_model_v1",
            "model_type": "probability",
            "prediction": 0.866, // model prediction (float)
            "updated_at": "2025-07-03T08:55:33.301099+00:00"
        },
        {
            "user": "p-478823",
            "model_name": "early_vip_model_v1",
            "model_type": "probability",
            "prediction": 0.0018,
            "updated_at": "2025-07-02T09:29:50.075361+00:00"
        }
        // additional predictions for other users
    ]
}

Explanations:

datetime : when was the callback sent.

data : array with predictions from models for users.

user : user id.

model_name : the model that made the prediction.

model_type : model type, probability means that the prediction has a float type and a value between 0 and 1.

prediction : prediction result (float).

updated_at : when was the prediction updated.

How to integrate callbacks

To begin receiving callbacks:

• Prepare a server that receives POST requests with JSON payload of the format specified above.
• Send us a URL to which we should send the callbacks.

After these steps we will configure the callback forwarding to your server.