Microsoft Case Study
Azure cost optimization for a SaaS company.
Nine month engagement. Reserved Instance coverage from 34 percent to 86 percent. Storage tier mix rebuilt. Renewal discount up two points against a larger commit.
Nine month engagement. Reserved Instance coverage from 34 percent to 86 percent. Storage tier mix rebuilt. Renewal discount up two points against a larger commit.
A growth stage SaaS company cut its Azure run rate by 31 percent in nine months while accelerating new product launches and tightening FinOps discipline across teams.
The client is a vertical SaaS platform serving the field services industry. The Azure footprint had grown from 600 thousand dollars in 2022 to 8.4 million dollars in 2025.
The CFO wanted a clean view of where the money was going. The CTO wanted a way to keep new product velocity without the bill running away.
Our team ran a nine month optimization program that combined a usage audit, a discount mechanics rework, and a FinOps operating model the company could run on its own afterwards.
The client serves more than 4,000 mid market customers across North America and the UK. The product is multi tenant. Each customer touches a shared Azure footprint that scales with bookings.
Engineering had grown from 45 to 220 people in three years. Cloud spend grew faster than headcount.
The 2025 board pack flagged cloud as the second largest cost line after payroll. Margins were under pressure from new ARR mix shift toward smaller customers.
The CFO wanted Azure spend back to a known ratio against revenue, not back to a fixed dollar number.
The team had bought a small block of one year reservations in 2023 and never expanded them. Almost two thirds of the compute hours were running at on demand rates.
On demand pricing for steady production workloads is roughly 60 percent more expensive than a three year reservation. The gap was funding nothing strategic.
Production database tier had been sized for peak Black Friday throughput from a 2022 launch. The actual peak CPU never crossed 28 percent in 2025.
Twenty five VMs accounted for 41 percent of the entire compute bill. Rightsizing those alone produced more savings than touching the rest of the estate.
Every nightly database backup created a premium SSD managed disk snapshot. No retention policy had ever been set. 38 terabytes of cold backups were sitting on the most expensive tier.
Cool blob storage was forty times cheaper for the same data. Moving the snapshots saved 41 thousand dollars per month with no operational impact.
Cross region replication had been turned on for a feature that shipped in 2023 and then deprecated. Traffic kept flowing and kept billing. Nobody owned the line item.
Azure run rate before and after, by category
| Category | Before (monthly) | After (monthly) | Change |
|---|---|---|---|
| Compute (VMs, App Service, AKS) | $418,000 | $262,000 | -37% |
| Database (SQL, Cosmos DB) | $104,000 | $71,000 | -32% |
| Storage (managed disks, blob) | $87,000 | $36,000 | -59% |
| Networking (egress, ER, gateways) | $48,000 | $29,000 | -40% |
| Other (Sentinel, Monitor, services) | $43,000 | $38,000 | -12% |
| Total Azure infrastructure | $700,000 | $436,000 | -38% |
We did not save thirty one percent by being clever. We saved it by giving engineers a number, a deadline, and a reason to care.
We rebuilt the Reserved Instance and Azure Savings Plan portfolio against a 36 month forecast. The mix is now 65 percent three year reservations, 20 percent one year, and 15 percent Savings Plan flexibility.
Engineering owned the rightsizing decisions. Finance owned the savings tracking. We facilitated the meeting where the two sides agreed the rules.
The rule was simple. Any VM running under 30 percent CPU for ninety days dropped two sizes. Any VM running over 70 percent jumped one size. Nobody argued after the first week.
A six line PowerShell script tagged every snapshot older than thirty days and moved it to Cool or Archive. A second script enforced the same retention going forward.
Every Azure resource now carries an owner team, a product area, and an environment tag. The monthly cost report goes to engineering team leads, not just to finance.
Within two months, three engineering teams deleted their own pre production environments after seeing the cost on their dashboard.
We sequenced the optimization work so that the new run rate was visible by the time Microsoft started talking about the EA renewal.
Microsoft offered a deeper discount band, not a shallower one, because the forward commit was credible and the consumption pattern was clean.
Cloud cost is now a monthly review owned by engineering team leads. Finance reviews variance against the forecast, not the line items.
Every new product epic has an Azure cost estimate in the design doc. No epic ships without a forecast.
Roughly two thirds came from engineering and discount mechanics work. About one third came from the renewal posture and the way Microsoft was briefed on the new run rate before the formal renewal opened.
No. Two new product modules shipped during the program. The cost forecast for those modules was built into the workstream from week one, so the savings target accounted for the new load.
Some companies do. The advantage of bringing in our team is speed and renewal leverage. The engineering work is well known. The Microsoft side requires experience to avoid leaving discount on the table.
Six to nine months from kickoff to a stable new run rate. The first 90 days deliver about sixty percent of the eventual savings. The remaining work locks in the operating model.
It varies by business model. SaaS platforms with multi tenant architectures typically target 8 to 12 percent. Single tenant or compute heavy products run higher. The right number is the one your CFO can defend at the next board meeting.
It locks in the discount, not the technology. Reservations apply at the subscription or management group level and can be exchanged or cancelled with a small fee. The flexibility is higher than most teams assume.
Microsoft renewal moves, the EA framework, the M365 SKU framework, the Copilot framework, and the buyer side moves across the full Microsoft estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
Once engineers saw the cost of their own pre production environments, three teams deleted theirs the same week.
500+ enterprise clients. 11 vendor practices. Industry recognized. One conversation can change what you pay for the next three years.
Buyer side Azure intelligence for CIOs, CFOs and FinOps leads. No spam. Unsubscribe anytime.
