You can use Adikteev Churn API to pull a list of your app users ranked by churn probability and leverage our machine learning capabilities for diverse marketing purposes. Feel free to reach out to your Adikteev client partner to know more about the possible use cases.

Prerequisites

Data stream

To start the churn prediction model, we will require access to your organic app events. You can refer to this documentation to activate this data stream from your MMP.

Once this step is completed, we will need a minimum of 7 days of data history to compute churn scores for your users. Churn predictions are recomputed daily and they reach their highest accuracy after 3 months or more of data history.

API credentials

To access the API, you will need to have a login/password that is shared during your subscription to the service.

First, your Adikteev client partner will send you an email to generate a password.

This password will be used to access the API with your email address. 

Using the Churn API

API authentication

Each API call must contain an HTTP authentication header, in the following format:

authentication: Basic mytoken

where mytoken is the string "your_email:your_password" encoded in Base64 format

For example, if your email is johndoe@company.com and your password is mypassword, the authentication header should be:

authentication: Basic am9obmRvZUBjb21wYW55LmNvbTpteXBhc3N3b3Jk

Base path

https://churn-api.adikteev.com

Available routes

/churn_scores

Mandatory query parameters: 

• bundle: app bundle identifier, in the format "com.domain.appname". Beware for iOS apps this is not the numerical store ID.
• os: "ios" or "android"

Optional query parameters:

• date: "YYYY-MM-DD" (or just "latest")

⚠️ Past churn scores are stored for 90 days, after that they are deleted and not accessible anymore 

API output

Each valid call to Adikteev Churn API will return a CSV file with the two following columns:

• IDFV for iOS apps or GAID for Android apps
• Churn bucket: Integer from 1 to 10 representing the predicted churn probability (1: highest probability - 10: lowest probability)

Sample output

idfv,churn_bucket
00000C78-A0CB-4C31-865B-F7633F92C64D,1
87F9D23D-8886-4262-B4C9-BBCC68F94262,1
1A720ABF-38A3-4807-A774-99F3E018B710,3
87F9C6E6-E25B-4F98-821C-B89F3B3BF02F,8
87F9BD65-21B9-4AEF-87F1-6AF141B72940,1
87F8FA1A-FB2D-4633-899B-23EE574C9817,10

Examples using curl

- without date will return the latest available churn scores:

curl -L --output "myscores.csv" --header "authorization: Basic c29tZF91a2VyLnNv1WVfcGFzc3dvcmQ=" \
  'https://churn-api.adikteev.com/churn_scores?bundle=some_bundle&os=some_os'

curl -L --output "myscores.csv" --header "authorization: Basic c29tZF91a2VyLnNv1WVfcGFzc3dvcmQ=" \
  'https://churn-api.adikteev.com/churn_scores?bundle=some_bundle&os=some_os&date=latest'

curl -L --output "myscores.csv" --header "authorization: Basic c29tZF91a2VyLnNv1WVfcGFzc3dvcmQ=" \
  'https://churn-api.adikteev.com/churn_scores?bundle=some_bundle&os=some_os&date=YYYY-MM-DD'