But these aren't the right services where the test should be, right?
There's another service that says "ok we take the 100 bytes from A, and we take the $17 SKU from B, and this should equal $X".
It's the third service that multiplies these things that failed. Where are the tests for that?
To me this sounds like a human-or-LLM-driven error. There must be a pretty limited set of factors that determine a pricing unit: I’m not really sure how a deterministic system could do that infrequently enough to not be a bigger story. Maybe a reeeeaaaallly rare race condition or something like that? To me this smells like having enough manual work involved in the process to fuck something up, but not nearly enough eyes on it to notice.
I’m totally guessing though.