One incident, many symptoms
When a hosting or network provider has an incident, the visible symptom is usually somebody else's website failing. A Cloudflare problem shows up as error pages on unrelated sites; a Vercel or Netlify build incident shows up as a deploy that never finishes. Checking the provider directly is faster than checking each affected site.
Cloud status pages are large. They cover dozens of regions and products, and an incident in any of them raises the overall state. Before assuming your own workload is affected, open the service page and read the incident title, which normally names the region or product.
Control plane and data plane
Providers distinguish between the systems you use to manage resources (dashboard, API, builds) and the systems that serve your traffic. An incident in the first group stops you from making changes while what is already deployed keeps running. The component list on each service page shows which side is declared as affected.