Why is our cloud bill higher after migration than the data centre was?

This is common enough to be predictable. It is almost never the compute rate — it is what got carried across unchanged.

Cause 1 — provisioned for peak, running at idle

On-premise you sized for the busiest hour and paid once. In cloud you pay for that sizing every hour. Instances running at low utilisation around the clock are usually the single biggest line. Right-size against observed usage, not the original spec.

Cause 2 — no lifecycle on storage

Backups, snapshots and logs accumulate because nothing deletes them. Storage looks cheap per unit and becomes significant in aggregate. Lifecycle policies that tier and expire old objects usually pay for themselves within a month.

Cause 3 — data transfer between zones

Chatty services placed in different availability zones generate transfer charges that did not exist on a flat internal network. Colocate services that talk constantly, and check that your load balancing is not routing traffic across zones by default.

Cause 4 — everything on demand

Baseline workloads that run continuously should not be on on-demand pricing. Committed use discounts on the steady portion of your estate are the least disruptive saving available.

Where to start

Tag everything by owner and environment first. You cannot reduce a bill you cannot attribute. Most organisations find that a small number of untagged, unowned resources account for a surprising share of the total.