This whole thing has that eerie feeling of a brand signal being choked off before it ever reaches Meta's optimisation engine. GTM is fine for housing the Pixel - it's all about clean triggering and deduplication. honestly, the real culprit here is almost certainly the CallRail integration.
what stands out to me is that the errors are clustering exclusively around CallRail call events, and the custom conversions aren't registering high-value calls. That smells less like a broken pixel and more like either a naming mismatch in the event parameters, missing required fields in the API payload, or the timing being off - Meta basically sees a weak, invalid signal and doesn't bother to optimise against it.
And yes, CallRail's API-driven events are a nightmare because you don't control how the payload gets structured before Meta receives it. Everything can look technically connected on the backend, but the optimisation layer still treats those call events as noise. It's one of those maddening situations where the whole brand funnel looks healthy on paper, but the underlying signal is essentially garbage
