Azure has been mainstream in UK enterprise IT for long enough now that most organisations are past the “should we go to the cloud” stage. The question is more often about what’s already running in Azure, whether it’s running efficiently, and whether the original business case actually held up. And on that last point, a lot of boards are quietly discovering the answer is no.
The pattern is familiar. A cloud migration was approved three or four years ago on a spreadsheet that promised lower costs and greater agility. The migration happened, more or less. The costs did not go down. In many cases they went up, sometimes by a lot. Nobody wants to revisit the original business case in front of the audit committee, so the conversation drifts sideways into discussions about “optimisation” without anyone saying out loud that the last round of consultants oversold the savings.
Lift-and-shift was always a half-measure
A large proportion of UK Azure estates were built by lifting virtual machines out of on-premises data centres and dropping them into Azure more or less unchanged. This is sometimes called lift-and-shift, and it was sold as a pragmatic first step. The reality is that it delivered most of the downsides of cloud (variable billing, new skills needed, unfamiliar tooling) and relatively few of the upsides.
The upsides of Azure, when they arrive, come from using the platform as a platform. That means platform-as-a-service offerings instead of VMs, managed databases instead of SQL Server running on a Windows box, event-driven architectures instead of scheduled jobs, and serverless compute where it fits. It also means getting serious about Infrastructure as Code, so that the environment can be rebuilt, audited, and version-controlled rather than being a collection of clicks in the portal that nobody remembers making.
Organisations that only completed the lift-and-shift step often end up in a worse position than they started. They have the ongoing cost of Azure plus the residual cost of operating systems, patching, and licensing that they thought they were leaving behind. The migration technically happened. The modernisation did not.
The cost conversation nobody wants to have
Azure spend is one of those line items that grows quietly until someone notices it. There are a few specific patterns that show up repeatedly in UK organisations.
The first is overprovisioning. Virtual machines sized for peak load that now run at 4% average utilisation because someone was nervous during the migration and specced up to be safe. The second is storage sprawl, where snapshots and backups from 2022 are still quietly accruing charges because nobody set a retention policy. The third, and probably the most common, is a lack of any structured FinOps practice, which means nobody owns the monthly bill, nobody tags resources consistently, and nobody can say with confidence which application costs what to run.
None of these are Azure’s fault. They’re all fixable. But they rarely get fixed without deliberate effort, and that effort usually requires someone outside the incumbent IT team who can ask uncomfortable questions without being politically invested in the answers.
Why the Expert MSP designation actually matters
Microsoft maintains a tier called Azure Expert MSP that sits above its standard partner designations. The distinction is meaningful because the audit is done by a third party, not by Microsoft itself or by the partner, and it covers operational practices rather than sales performance. There are only a few dozen Azure Expert MSPs worldwide, and fewer still in the UK.
For organisations looking at serious Azure work, whether that’s a migration, a modernisation programme, or a cost-reduction exercise, this tier is a useful filter. One UK Expert MSP’s outline of its Azure consulting services covers readiness assessments, migration planning, and the Azure Migration and Modernisation Program funding that Microsoft makes available through qualifying partners. It’s a reasonable starting point for understanding what to ask for in a statement of work. Non-Expert partners can absolutely do good work, but the Expert tier correlates strongly with the operational discipline that separates a smooth migration from a painful one.
What good looks like in 2026
A well-run Azure estate in 2026 has a few observable characteristics. Spend is tagged and attributable. Reserved instances and savings plans are used where they make sense. Workloads are on the right services rather than just on VMs by default. Security posture is monitored through Defender for Cloud rather than assumed. There is a documented landing zone architecture that new workloads have to fit into, rather than a free-for-all where every team stands up its own resource group with its own naming convention.
Organisations don’t get to that state by accident. They get there by treating Azure as something that needs continuous stewardship, not as something that was “done” when the last workload came out of the data centre. The boards that are happiest with their Azure investments are the ones that have accepted this and built the operating model to match. The ones still surprised by the monthly bill are usually still pretending the migration was the finish line.
Leave a Reply