Graceful Degradation
Continuing to deliver reduced but useful function when a dependency fails, instead of failing the whole request.
Every dependency should have a documented answer to "what does this system do when you are gone?" If the answer is "returns a 500", the dependency is critical and the architecture has inherited its availability.
The alternatives are almost always available and almost always cheaper than making the dependency more reliable: serve a stale cached value, hide the personalised section and show the generic one, accept the write into a queue and process it later, drop the enrichment and return the core record.
Doing this well requires deciding which parts of a response are essential and which are enhancements — a product conversation as much as a technical one, and one worth having before the incident rather than during it.