Big tech companies move slow technically, but move fast in organization structure. A "long term" project often means nothing ever gets done, a management chain change happens, and it all goes to nothing.
Metrics are never perfect, but it beats hand wavy claims of success, normally if a change is useful there is some metric that will show it. Honestly, the default "success" in big tech is the scope being reduced so many times that literally nothing happens and the PM's and management treat it like a win.
Finally testing... It's already easy enough to get nothing done in a big tech company but test code can take this to another level. You can join a team, write a big design document about some testing framework, spend 6 months implementing it, then leave the team just in time for the next reinvention of the test framework. And the whole time you were never at risk of deploying anything to production.