Dated starting points only — Unknown until verified means no invented $/GB. PipeToll does not scrape private consoles or guarantee figures.
Path-class vocabulary: path template, Unknown until verified. Azure often says bandwidth; AWS says data transfer out. Normalize vocabulary before arithmetic. PipeToll keeps Honest Unknown cells rather than inventing list prices.
Path templates to model
Client ← Internet ← Blob/S3 — public download; zone/region free grants matter.
VNet/VPC compute ← private endpoint — private path fees replace some Internet meters.
GRS/CRR replication — background meters distinct from user reads.
CDN / Front Door / CloudFront origin — two-bill risk on miss.
Archive rehydrate then download — restore SKU + bandwidth/egress.
Common traps
Comparing only “Internet $/GB” ignores private endpoint fees and replication. Archive restores surprise teams that budgeted egress alone. Cross-cloud copy bills the source. No scorecrowns across Azure vs AWS from partial public pages.
FinOps checklist
Translate Azure bandwidth labels to the same path classes used for S3.
Split replication, private, Internet, and CDN-origin lines.
Include restore/rehydration when cold tiers are in play.
Verify rates with dated pricing URLs; else badge Unknown.
Worked example (educational): 2 TB Internet downloads + archive restore
Restore 500 GB from archive/cool tier (rehydration/restore SKU — often outside pure egress) then download 2,000 GB to Internet. Model restore and bandwidth/egress as separate lines; GB×$/GB alone undercounts. Placeholder rates stay Unknown until dated. Not a savings guarantee.