Cloud cost optimization is usually run as a finance exercise. Finance can see cloud spend. Engineering can change it. FinOps lives in the gap, and the gap isn't a reporting gap. It's a product gap.
THE CLOUD COST METRIC NOBODY OWNS
A monthly invoice isn't a metric anyone can act on. Cost per tenant, per request, per model call is. Once you name the unit, the number stops being a bill and starts being a design constraint.
TREAT THE CLOUD COST DASHBOARD AS A PRODUCT
I learned this the slow way at Deloitte. The same underlying billing data had to answer different questions for three audiences: leadership wanted spend against budget, finance wanted allocation by cost centre, engineering wanted to know which instances were actually idle. One combined report satisfied nobody. Splitting it into three views of the same data, each cut to the decision that group was making, is what actually got used.
CLOUD COST OPTIMIZATION: WHAT TO TRY FIRST
Pick one unit metric and publish its definition.
Give it an owner who can change the system, not just report on it.
Review it in the same meeting where roadmap trade-offs already happen.