I've been deep in SFMC for years, and yeah, the black box nature of email delivery is a constant headache—especially when you're chaining journeys and automations that depend on timely sends. Benchmarks like this are gold because they cut through the vendor hype and give you real metrics on latency, throughput, and reliability under load. I've lost count of the times a "high deliverability" promise turned into a nightmare when scaling up for a big campaign.

One thing I'd flag from my experience: don't just chase raw speed. Look at error handling and retry logic in these APIs. SFMC integrations often choke on transient failures, like rate limits or auth hiccups, so providers that bake in robust fallbacks save you from midnight alerts. Also, factor in regional performance—I've seen US-East endpoints fly while EU ones lag, which murders A/B tests across time zones.

If you're building something notification-heavy, I'd dig into the consistency of webhooks too; polling APIs feels prehistoric when you're trying to trigger real-time automations. This data helps narrow it down, but always run your own load tests with your payload sizes—email with heavy personalization or attachments behaves differently. Props for putting objective numbers out there; it's rare in this space.