DFdom@fradley
← ~/writing
$ cat strategy-before-the-statement-of-work.md

Write the cloud strategy before the MSP's statement of work

Pick up the statement of work for most Azure engagements and the first page names a deliverable, usually a landing zone or a number of servers to migrate. An MSP can price that and deliver it.

What's rarely on that page is why the organisation is in Azure at all, or who on your side gets the final say when two goals pull against each other. Those questions still come up during the build: which regions, how much resilience, whether policy blocks deployments or just reports them. With nothing written down, whoever is on the call picks a sensible default.

Flexera's 2026 State of the Cloud report, published in March from a survey of 753 cloud decision-makers, found 64% of organisations now use value delivered to business units as their main measure of cloud progress, up twelve points in a year. Wasted spend went up too, to 29%, the first rise in five years, and Flexera puts the rise down to surging AI workloads. When the board asks what the Azure spend bought, you want an answer that was agreed before the build.

Microsoft's Cloud Adoption Framework (CAF) puts Strategy two phases ahead of Ready, which is where the landing zone sits. The Strategy methodology was last revised in May 2026 and has four steps, but you don't need all of it before you talk to an MSP. I'd write one page covering the decisions a supplier can't make for you, and pull CAF in where it helps fill the gaps.

the customer writes this most statements of work start here Strategy why, and who decides Plan estate and skills Ready the landing zone Adopt migrate, build Govern · Secure · Manage run across all four phases Strategy's four steps: motivations and objectives, strategy team, preparing the organisation, then cost, resiliency, security and sustainability.
Microsoft's Cloud Adoption Framework in outline. A statement of work that opens with "build a landing zone" starts at Ready, two phases after the part only the customer can write.

The brief

Seven sections, one page. Every row is the customer's decision.

CLOUD STRATEGY BRIEF, for [MSP name], version [n], [date]

1. Why we are in Azure
   Motivation              Priority   Urgency   Notes
   [e.g. resiliency]       High       Medium
   [e.g. datacentre exit]  Medium     High      contract ends [date]

2. Mission (one sentence)
   [What cloud is for in this organisation]

3. Objectives and key results
   Objective   Key result (measurable, dated)   Owner (our side)   Review
   [ ]         [ ]                              [name, role]       quarterly

4. Who decides
   Strategy team: [IT lead], [finance lead], [security lead], [sponsor]
   Design exceptions approved by: [name]
   Budget changes approved by: [name]

5. Design positions we have already taken
   Resilience tier by workload class: [ ]
   Policy enforcement: audit until [date], then deny
   Regions: [primary], [secondary if any]

6. What we are not doing
   [e.g. no second cloud provider; no re-architecture during the exit]

7. After the engagement
   Platform owner (our side): [name]
   Run model: [internal | managed service | split, with the split written down]
   Handover criteria: [ ]
Brief section CAF Strategy step Where the MSP comes in
1 Why we are in Azure 1 Motivations Asks about it in discovery
2 Mission 1 Mission Designs against it
3 Objectives and key results 1 Objectives Works towards them
4 Who decides 2 Strategy team Knows who to ask
5 Design positions 4 Inform the strategy Prices each option
6 What we are not doing (yours alone) Quotes for less
7 After the engagement 3 Prepare the organisation Runs it, if you choose

CAF has you classify each motivation by priority and urgency in separate columns. A datacentre exit is urgent, but that says nothing about its priority. Without the second column it can end up driving the whole strategy.

Keep the mission to one sentence about what cloud is for in your organisation. CAF's own example, "Use cloud technologies that align with our business and IT goals to empower our organization to innovate, scale, and deliver secure, high-quality services", would fit almost any company, so it gives an MSP little to design against. Its example key results are more useful: a 50% reduction in P1 incidents within a quarter, or infrastructure spend down 20% within twelve months, each with an accountable role and a review, "preferably quarterly". Put one of your own people's names against each before the MSP starts.

For the strategy team, CAF says start small: "Core members at this stage typically include IT, finance, and security." I'd make sure finance is one of them, mainly for the moment the MSP asks whether a workload should run across availability zones (with the extra instances that implies). A finance lead who has seen the objectives can answer that in the meeting instead of finding out from the invoice.

Don't worry if section 5 is half empty to begin with. The MSP can help here by pricing each resilience tier, region and policy setting before you pick one. Audit mode is easier during a migration; a compliance motivation usually means deny. Fill in section 6 even if it's one line, because every exclusion is scope that comes off the quote.

An MSP build is a project, in CAF's terms, and when projects finish, "the ownership often shifts to another team for operations and maintenance." So before the build starts, name your platform owner and decide whether the MSP runs the platform after handover. CAF also recommends regular reviews of how partners contribute to your key results. The written brief lets the MSP's service review report against them alongside its SLA.

Five questions for the first meeting

Expect the MSP to ask for most of this in discovery anyway. If you bring it, the first meeting can start on design. Ask:

  1. Which of our key results does each workstream in your proposal support? Challenge any that supports none.
  2. Which design decisions will you make by default, and which do you need from us? Get the list in writing, with region, network topology, resilience tier and policy enforcement mode on it.
  3. What does handover look like, and what do you need from our platform owner before then?
  4. How will you report against our objectives, and how often?
  5. What would you advise us not to do?

The methodology overview has a checklist with separate startup and enterprise tracks. If the migration question in section 1 is the sticking point, Lift-and-shift with a written end date goes through rehosting versus re-architecting, workload by workload.

discuss on linkedin → more writing