What a cloud migration really costs

Cloud migrations are usually justified on infrastructure cost. That is the weakest part of the case, and the part most likely to disappoint.

What the business case usually says

Hardware refresh avoided, data centre space released, some licensing consolidated. All real, all easy to total up, and all roughly cancelled out in year one by the cost of getting there.

What it usually misses

The migration itself. Someone has to do it, and that someone is usually the team already running everything else. Migration work is not free just because it is internal.

Two environments at once. You pay for the old and the new until the last system moves. That overlap is routinely underestimated, and it grows every time a cutover slips.

Drift. Cloud spending grows quietly. Resources spun up for a test stay running. Storage tiers are never reviewed. Nothing is visible until the invoice arrives, and by then it is a year of habit.

Where the return actually comes from

Not the hosting line. It comes from the number of things you can now try. When an environment takes minutes instead of weeks, teams experiment, because being wrong is cheap. That changes what gets built, which is worth considerably more than a modest saving on compute.

For a mid-sized company, reaching a customer two months earlier usually dwarfs the infrastructure difference either way.

Keeping it under control

Treat cost as an engineering concern rather than a finance one. Spending attributed to the team that creates it, budgets visible to the people making the decisions, and a regular look at what is running and why. Done from the start this is straightforward. Introduced after two years of drift, it is a project of its own.

Choosing a path per system

Not everything deserves the same treatment. Move stable systems unchanged. Re-platform the ones where a managed database or a container removes real operational work. Rebuild only what is central to the business and genuinely held back by its current design. Most estates need all three — the mistake is picking one strategy because it is easier to explain.