Starting with one is entirely reasonable: metrics and logs are the typical entry point because they require minimal instrumentation overhead. Without traces, however, distributed root-cause analysis in microservices requires manual inference across logs from multiple services. Most mature observability programs converge on all four signals because each fills a gap the others leave.