Since Replicate is generally used to load to the target in near to real-time , it is not advisable to try to execute every time Replicate loads data into the target.
Rather, the best approach is to define a workflow (if needed to run the CDC ETL and then Data Mart), and schedule that task to run every X. For example every 15 minutes. When executing, Compose will detect via the _CT tables if there is data to be loaded and only process tables that have data waiting to be processed. (This is via the CDC tasks).
Note, that if you are using Replicate for FULL LOAD only - i.e. to load a file to your DW, then instead of scheduling in Replicate, you can assign a Landing area in Compose to that task, and have Compose orchestrate the Replicate full load job, followed by the DW / DM ETL processes via a Workflow. This is ONLY available for full load tasks as they have a discrete START and STOP (whereas CDC tasks in Replicate are continuously running).
Thanks everyone for your suggestions. I got the information I was looking for, which is that we cannot do a real time replication in Qlik Compose like we can do in Qlik Replicate. So, I scheduled my project to refresh data every 15 mins.