Is Hybrid ECS worth migrating to for render performance?

I don’t remember if, or which samples may have that, but you can look up into Unity ECS samples on github. There is heloCube repo. Maybe it contains some good reference point. But usually these samples are using hybrid rendering API.

Depending of the camera angle and movement, in some scenarios combination of dot product vectors, detecting if objects are behind camera and distance to object, may be sufficient. Or even just distance (top down cam). In other cases, you need some frustum solution, which is run on jobs.

Also, you could investigate one @DreamingImLatios framework, which has own thread on Unity ECS forum.