When one user request touches five services, logs alone cannot tell you which hop was slow or which one returned a 500. A trace stitches all the hops into a single timeline using a shared trace ID, so you can see exactly where latency or errors appear. Without tracing you end up grepping logs across services trying to correlate timestamps.