What it shows well
- Physical damage to fixed infrastructure (buildings, airfields, ports).
- Vehicle and aircraft presence/absence at named locations.
- Construction activity over multi-week windows.
- Crater patterns consistent with specific munition types.
What it shows poorly
- Anything indoors.
- Anything that moved between passes.
- Anything obscured by cloud, smoke, or active operations.
- Casualty counts of any kind.
- Intent.
How to read a published comparison
- Note the dates of the "before" and "after" frames. A "before" frame from weeks earlier may show changes unrelated to the claimed event.
- Note the imagery provider. Different providers offer different resolutions and revisit cadences.
- Look at the published reasoning. Strong OSINT will mark exactly which features support the conclusion.
- Remember that the conclusion is usually narrower than the headline. "Damage consistent with a strike" is not "damage caused by this specific actor".