Satellite data is free, enormous, and intimidating. Most students who want to use it stall at the same three points, and all three are avoidable.
Pick a small area and a short window
The instinct is to download everything. Global archives are measured in terabytes, and a full pull will exhaust your patience before you learn anything. Choose one county and one growing season. You can always widen scope after the analysis works.
Understand the processing level
Earth observation products come in levels. Raw instrument readings are not the same as calibrated measurements, which are not the same as gridded, gap-filled products. Beginners often analyze raw data and conclude the satellite is broken. Start with the highest processing level that answers your question.
Expect gaps and know why they exist
Clouds block optical sensors. Orbits leave coverage holes. Frozen ground and dense canopy interfere with soil moisture retrieval. Missing values are not corruption, they are physics, and handling them explicitly is part of the work.
Validate against something on the ground
Wherever possible, compare your satellite-derived numbers to a ground station. The comparison rarely matches perfectly, and understanding why is where the actual learning happens.
A first project that covers one county for one season, validated against one weather station, teaches more than a global analysis nobody can check.