“Sub-metre” gets used as a synonym for “good,” but it has a precise meaning, and understanding it tells you exactly what kind of decisions an image can support.
Resolution, plainly
Ground sample distance (GSD) is the real-world size of a single pixel. At 10-metre resolution — typical of free, frequent imagery like Sentinel-2 — one pixel covers a 10×10 metre patch. You can see fields, large buildings, and broad land-cover change, but not a roof's condition. At 0.5-metre GSD, a pixel covers half a metre, and individual structures, vehicles, and roof detail come into focus.
The question is never “is this high resolution?” but “is this resolution enough for the decision I am making?”
Resolution versus revisit
There is a trade-off most people miss. Coarse imagery is often free and captured every few days, which makes it excellent for detecting change over time. Sub-metre imagery is licensed and captured on demand, which makes it the right tool for confirming detail. Strong intelligence usually combines them: frequent passes flag that something changed, and a high-resolution capture shows exactly what.
What sub-metre does not do
It does not read licence plates or identify people — that is the realm of fiction, not commercial Earth observation. It also cannot recover detail that was never captured: upscaling a coarse image does not create real resolution, it invents pixels. Honest sourcing matters, which is why a report should always state where its imagery came from and at what resolution.
Used well, sub-metre imagery is the difference between knowing a property changed and knowing precisely how. That precision is what turns overhead data into a decision.