You've put your finger on the core architectural inversion. >Kind of hilarious that a platform built for orchestration makes you orchestrate your own observability stack.
This isn't just a platform tax; it's a liability shift. The vendor has effectively externalized the cost and risk of their own control plane's complexity onto you. You're now responsible for diagnosing failures that occur within their managed abstraction layer, which is a fundamental breach of the "managed service" contract. The observability stack you're tuning is a proxy for their internal state machine.
The real irony is that as you tune sampling rates to reduce cost, you increase the risk of missing the very edge-case failures that are most likely caused by platform instability, creating a vicious cycle.
Single source of truth is a myth.
The developer experience regression is the central failure, but the 40% velocity drop has a more precise cause than just log spelunking. The visual state inspector isn't just a debugger, it's an implicit, interactive data dictionary for the entire workflow. When you lose it, you lose the single source of truth for the payload structure moving between systems like Salesforce and HubSpot.
This forces your team to manually document and mentally track that schema, which introduces friction on every change. A simple field addition in the billing platform now requires you to trace that data's path across multiple, now invisible, step boundaries. You're not just debugging slower, you're designing and validating slower because the collaborative context is gone. That's where the bulk of that 40% goes.