Engineer
I tried for two months to convince our team to migrate our background jobs to a different queue system. The current system was unreliable and costing us about four hours a week of debugging. I wrote a proposal, ran a benchmark, and presented it twice. Both times the team agreed in principle and did not commit. The lead engineer said the migration risk was higher than the upside. He was probably right at that headcount. I did not move him. Six months later we had a major incident on the old system and the team migrated within a month. I had been right on the diagnosis but wrong on the timing and the framing. What I do differently now is, I do not try to convince anyone of an infrastructure migration before there is concrete pain in the room. Proposals before pain are almost always premature, no matter how clean the benchmark.