What cases or conditions determine whether I should use incremental load or full load? We can do a full refresh every day, isn't it? How do I decide between incremental and full load?

Key conditions include: 1) Data freshness needs – if near real‑time is required, incremental is essential; 2) Source change detection – if the source provides a change flag or timestamp, incremental is feasible; 3) Data volume – large tables benefit from incremental to avoid long runtimes; 4) Cost and resource limits – incremental uses fewer compute and storage; 5) Schema stability – if the schema changes often, a full load may be safer. If you can tolerate a daily full refresh and the dataset is small, a full load simplifies the pipeline. For larger or rapidly changing data, incremental reduces load time and network traffic. A common strategy is to start with a full load, then switch to incremental once CDC or change tracking is available.