Reading the same data from a slow database again and again wastes time and load. A cache keeps a copy in something faster, like Redis or a CDN, so most reads skip the slow path. Reach for it when the same data is read far more often than it changes, such as product pages or user profiles. Caching brings stale-data problems, so keep TTLs short for fast-changing data, and avoid it where you must always show the latest value.