The Azure migration playbook that procurement will accept

Why migration plans fail

Most Azure migration plans fail not in the technical architecture — they fail in the cost model. A plan that cannot survive a CFO's scrutiny will not survive procurement. And a plan that cannot survive procurement does not ship.

The three most common failure modes are:

1. The cost model is a point estimate, not a range. Procurement reviewers have seen enough migration projects to know that point estimates are fiction. A range with assumptions stated explicitly is far more credible than a precise number with no backing.

2. The FinOps layer is missing entirely. Most migration plans stop at "lift and shift the workloads." The question no one asks in the planning phase — and everyone asks six months after go-live — is: why is the Azure bill 40% higher than the model predicted? The answer is almost always licensing, reserved instance choices, and unattended resources.

3. Risk is buried in an appendix. Procurement reviewers read the executive summary and the risk section. If the risk register is a boilerplate list of generic technical risks with no mitigation plans, it signals that the team has not thought through the hard parts.

What a procurement-ready migration plan contains

A plan you can take to Treasury Board, a CFO, or a Big Five bank's vendor review should contain six elements:

Target architecture — a diagram that a non-technical reviewer can follow, with the current state and the target state clearly labeled and a narrative that explains why each choice was made.

Cost model with ranges — a three-column model (low, expected, high) with assumptions for each key driver: compute, storage, licensing, egress, support, and internal labor. The FinOps overlay — reserved instances, Azure Hybrid Benefit, right-sizing — belongs here.

Sequencing plan — which workloads move first, why, and what dependencies exist. The dependency map is often the most valuable artifact in the plan because it surfaces scheduling risks that cannot be seen in the architecture diagram alone.

Risk register — genuine risks, honestly assessed, with mitigation plans that are specific and owned by named roles. Not a boilerplate list.

TCO comparison — the three-year total cost of the current state versus the migrated state, including infrastructure, licensing, and operational labor. This is the number that most justifies the investment.

Governance and compliance summary — for public sector clients or regulated industries, a one-page summary of how the architecture meets applicable compliance requirements (PBMM, PIPEDA, provincial privacy regimes, OSFI guidelines as relevant).

The FinOps layer in practice

Reserved instances and Azure Hybrid Benefit are the two largest cost levers in most migrations. Organizations that plan their reserved instance strategy at migration time — rather than retroactively — typically reduce their steady-state Azure bill by 20–35% compared to pay-as-you-go.

The catch is that reserved instance commitments require confidence in the workload sizing. That is why a migration sprint that includes a right-sizing exercise — before committing to any reservation — pays for itself within the first three months of operation.