Ask HN: How to calculate SaaS COGS for production environment when DevOps

Hi all, Curious to know how folks have been accounting for COGS for a SaaS service in an environment where the engineers responsible for maintaining production rarely actually touch production directly.

In the olden days, we would have a few engineers dedicated to managing production systems, and we would just use their time in the calculations. That was fine, because they spent much of their time looking at the bits, sometimes changing systems, and generally just being responsible for prod.

My current system is in the newish model where nobody actually touches production. All changes are software driven, logs are analyzed automatically with monitors, and generally only investigated when issues arise (detected or reported).

Anyone have a rationale for how to allocate the infrastructure teams time to production for this accounting exercise? I'm guestimating 50-50 split, prod verse non prod support (developer support), but I'd love to hear other POVs. Thx.