Help: Our SASE solution can't seem to handle UDP-based VoIP traffic reliably.

Hey everyone, I'm hitting a real roadblock with our SASE rollout and hoping someone has walked this path already.

We recently consolidated onto a major SASE platform (leaving names out, but it's one of the big three). Everything's been great for web traffic and SaaS apps, but our VoIP calls—which are mission-critical for our sales team—are suffering. We're using a UDP-based system, and the audio quality is all over the place: choppy audio, drops, and sometimes one-way audio. It's like the packets are just getting lost or massively delayed once they hit the SASE cloud. Our network team says the raw bandwidth is fine, so I'm convinced it's something in the SASE policy or architecture.

Here's what we've tried so far:
* Created a dedicated traffic steering rule to send VoIP traffic directly to the nearest SASE PoP, bypassing some of the deeper inspection.
* Played with the "loss tolerance" and "latency sensitivity" settings in the policy (where available).
* Ensured QoS markings (DSCP EF/CS5) are preserved from our internal network.

The problem *feels* like it's in the secure web gateway or firewall-as-a-service layer. Maybe it's treating the constant small UDP packets as suspicious? Or the PoP selection isn't sticky enough mid-call?

Has anyone else fought this battle? Specifically:
* Did you have to create a completely separate, bypass policy for UDP VoIP traffic?
* Are there specific UDP timeout or session settings that made a difference?
* Any lessons on how you convinced the security team that a bypass was necessary for this specific protocol?

I'd love to hear any real-world config snippets or migration stories. Our sales ops lead is breathing down my neck because lead scoring depends on those call recordings being clear!

~jennam