First page of Microsoft's 100,000-partner directory, sorted by responsiveness Microsoft Solutions Partner — Security, Modern Work, Infrastructure, App Innovation Microsoft partner since 2006 1,100+ organizations under management
Home/Services/Managed Azure FinOps and Cost Guardrails Service
Managed ServicesConsulting

Managed Azure FinOps and Cost Guardrails Service

Managed Azure FinOps and Cost Guardrails Service is a recurring monthly cost practice for an Azure estate. Each cycle IT Partner maintains your cost allocation model and tagging hygiene, keeps budgets and daily anomaly alerts configured in Microsoft Cost Management, runs a waste review (idle, orphaned, oversized and unattached resources, non-production schedules, storage tiering and log retention), tracks reservation and savings-plan coverage and utilization with purchase, re-scope, exchange or trade-in recommendations, checks Azure Hybrid Benefit and the entitlement behind it, maintains a showback and unit-cost view, produces a savings-actions list executed with your approval or routed to your change control, and delivers an executive report that states what was actioned and what it changed. It costs $495 per month for the subscriptions and the monthly Azure spend band agreed in your order; larger, multi-tenant or unusually fragmented estates are quoted per estate. Savings are measured and reported, never guaranteed and never billed as a share of them — your Azure consumption and any reservation or savings-plan purchases stay yours, on Microsoft's meters. The one-time deep assessment, resource-health and patching monitoring, and architecture rework are separate services this page names and links.

Timeline 30 daysService owner Roman SotnikMicrosoft AzureMicrosoft Cost ManagementAzure Advisor

What this engagement is

Azure bills rarely grow because of one bad decision. They grow by accretion: a proof of concept nobody tore down, managed disks orphaned by a VM rebuild, a Log Analytics workspace left on default retention, a non-production environment that runs 168 hours a week to be used for forty, a VM family reserved two years ago that nothing runs on any more, and a dozen resources whose owner has left the company. Nobody was careless; nobody owns the number either. Finance sees one line item that moves every month and cannot attribute it to a team or an application. IT sees an environment it was told to keep available. The usual response is a one-off cleanup, which works — and then drifts back inside two quarters, because nothing changed about who looks at the number next month. This service is the part that keeps it from drifting back. Every cycle has the same named artifacts: an allocation model and a tag standard with measured coverage; budgets and anomaly alerts that are actually configured, addressed to people by name, and triaged when they fire; a waste review that names resources rather than categories; a commitment position (reservation and savings-plan coverage, utilization, expiry dates) that is reviewed before renewal dates rather than after them; an Azure Hybrid Benefit and entitlement check; a showback and unit-cost view your finance lead can read without a translator; a savings-actions list with owners and change windows; and a report that says what was done and what it changed. The engine is Microsoft's own — cost analysis, budgets, cost anomaly detection, scheduled exports (including the FOCUS dataset) and Azure Advisor inside Microsoft Cost Management — so there is no third-party FinOps platform to license and no agent to install in your estate. IT Partner is a Microsoft direct-bill CSP partner, which matters more than it sounds here. For customers whose subscriptions sit on the Azure plan under a Microsoft Customer Agreement, Cost Management is natively available to us at customer scope in our own partner tenant — costs by customer, subscription, resource group, resource, service and meter, with budgets and reservation utilization alert rules we can create centrally and address to your people. We work equally well inside your own Microsoft Customer Agreement or Enterprise Agreement with the roles you delegate. Either way, one of the first things we check is whether your own team can see their costs at all: on CSP, the cost-visibility policy that lets subscription users view Azure charges at pay-as-you-go rates in their own tenant is off by default until a partner enables it, and plenty of estates have been running for years with the people who create the spend unable to see it. Three rails hold this page together, and they are worth reading before the deliverables. First, the fee is fixed. It is never a percentage of your Azure bill and never a share of what we save, so the advice costs the same whether the right answer this month is a three-year reservation or 'delete these eleven disks and change nothing else'. Second, Microsoft's meters are yours: Azure consumption, reservation and savings-plan purchases and any Microsoft licensing are billed to you at Microsoft's rates and are never resold, capped or absorbed into this fee. Third, savings are reported, not promised. Every action carries an estimate and the method behind it before it runs, the monthly report states what actually changed against your prior run rate, and the number that eventually lands on your invoice is Microsoft's arithmetic rather than ours. We publish no savings percentages of our own and will not quote one on a call — a FinOps practice that leads with a percentage is selling the one number it cannot control.

Success criteria

01Within the first cycle every in-scope subscription has a budget with actual and forecast thresholds, an anomaly alert with named recipients, and an agreed tag standard — and you hold a written baseline of the last twelve months (or all available history) by subscription, resource group, service and meter.
02Cost becomes attributable: an agreed share of in-scope spend resolves to an owner, an environment and an application or cost centre, the coverage figure is measured every month against the target you set at onboarding, and the showback view is produced from that model rather than from a spreadsheet rebuilt by hand each month.
03Every anomaly alert and every budget threshold breach is triaged inside the agreed response window, with a written cause where one can be determined and a recorded decision where the answer is that the spend was expected.
04A savings-actions list exists every cycle, and each item carries an owner, an estimated monthly impact with the method behind it, a risk note and a change window — then is executed with your approval, scheduled, routed to your change control, or explicitly declined and recorded as declined.
05The commitment position is known rather than assumed: reservation and savings-plan coverage and utilization are tracked each cycle, low-utilization commitments are flagged in the month they drift, and no reservation, savings plan, auto-renew date or Software Assurance renewal arrives without at least one cycle of notice.
06Leadership can answer, from the monthly report alone, what Azure cost, how that compares with budget and forecast, why it moved, what was actioned and what the action changed — without asking IT to reconstruct it.
07The artifacts are yours and portable: tag standard, allocation model, budgets and alert rules, scheduled exports, the action register and every report live in your tenant or your storage and survive the engagement, so another provider or your own team can pick them up without translation.

What you receive

Onboarding and cost baseline (first cycle): we confirm scope (which subscriptions, management groups and billing scope), obtain least-privilege access, and build a baseline of the last twelve months or all available history by subscription, resource group, service and meter. It includes a first pass over Azure Advisor's cost recommendations, an inventory of existing reservations and savings plans with their coverage, utilization and expiry dates, a measurement of current tag coverage, an audit of whatever budgets and alerts already exist, and — on CSP — a check of whether the cost-visibility policy is enabled for your tenant. You get a short onboarding note: the top opportunities, the guardrails we propose to build, and what we could not see and why.
Allocation model and tagging hygiene: a minimum tag set agreed with you (typically owner, environment, and application or cost centre), tag coverage measured every cycle, untagged spend chased to a human rather than written off, and tag inheritance enabled in Cost Management where your scope supports it — it is available for Enterprise Agreement billing accounts, Microsoft Customer Agreement billing profiles, and Microsoft Partner Agreement subscriptions on the Azure plan, and applies subscription and resource-group tags to child resource usage records for the current month within about a day. Enforcement rules in Azure Policy are recommended here and, where you want them deployed and version-controlled, scoped as a separate project.
Budgets and anomaly guardrails: budgets per subscription, resource group or environment with actual and forecast thresholds and named recipients (Cost Management allows up to five thresholds and five email recipients on a single budget, and action groups at subscription and resource-group scope), plus cost anomaly alerts on every in-scope subscription. We tune the thresholds against your baseline so the alerts stay credible instead of becoming another ignored mailbox rule. Budgets notify; they do not stop spend — automated responses (a Logic App, a Function, a webhook behind an action group) are a build we quote separately.
Monthly waste review with named resources, not categories: idle virtual machines and machines stopped but still allocated, unattached managed disks, orphaned public IPs, network interfaces, snapshots and stale restore points, empty App Service plans, idle gateways and load balancers, over-provisioned AKS node pools, oversized VMs and databases judged from utilization data rather than a guess, storage tiering and lifecycle rules, Log Analytics retention and commitment tiers, non-production start/stop schedules, and dev/test pricing where you are entitled to it. Findings that need a workload owner's judgement are routed to that owner rather than actioned quietly.
Commitment coverage and utilization: reservation and savings-plan coverage against eligible usage, utilization tracked each cycle, and a recommendation to purchase, re-scope, exchange or trade in — with the trade-offs stated in writing, because commitments are the one part of Azure cost work that can lose you money. Where the billing scope supports it we run reservation utilization alert rules; as a CSP partner these are created centrally at customer scope from our partner tenant, with your people included as recipients. Expiry and auto-renew dates go on a calendar reviewed before, not after, they land.
Azure Hybrid Benefit and entitlement check: which Windows Server and SQL Server workloads claim Azure Hybrid Benefit, which could, and — just as important — which claim it without an active Software Assurance or qualifying subscription entitlement behind them, which is a compliance exposure rather than a saving. We report the position and the renewal dates; you or your licensing advisor confirm the entitlement, because that is a contractual question about your agreements, not a portal setting.
Showback and unit-cost view: spend by team, application and environment produced from the allocation model, plus one or two unit-cost measures agreed with you — cost per active user, per customer, per environment, per transaction — tracked month over month so a rising bill can be read as growth or as waste. One honest constraint: Cost Management's built-in cost allocation rules for splitting shared costs are supported on Enterprise Agreement and Microsoft Customer Agreement scopes, not on Microsoft Partner Agreement scopes, so for CSP estates the shared-cost split is produced in the report and the exported data rather than inside the portal.
Monthly savings-actions list, executed with your approval: every item carries an owner, an estimated monthly impact and the method used to estimate it, a risk note and a change window. Low-risk items inside the guardrails we agree at onboarding (deleting an unattached disk, applying a schedule, deleting an orphaned IP) are executed by us once you approve the list; anything touching production capacity, a workload owner's environment or your change process is packaged for that process and tracked until it closes or is formally declined.
Monthly executive report and quarterly re-baseline: one page for finance — spend against budget and forecast, what moved and why, actions taken and what they changed measured against the prior run rate, commitment coverage and utilization, tag coverage, and what is waiting on a decision — with the working detail behind it for IT. Each quarter we re-baseline: the commitment strategy ahead of renewal dates, the unit-cost trend, the budget figures themselves, and whether the guardrails are still catching what they were built to catch.
Engineering on call for what the practice spins off, and your data left portable: when a finding needs a project — re-platforming a workload, redesigning storage or backup, a landing-zone correction, automated shutdown tooling — IT Partner scopes it and quotes it fixed-price in writing, and you pay after you approve delivery. Where you want the cost data in your own store, we can deploy Microsoft's open-source FinOps toolkit (FinOps hubs and its Power BI reports) into your subscription as a quoted build; it is open-source tooling and the Azure resources it runs on are your consumption.

How the work unfolds

1. Onboarding, access and baseline (first cycle)

We agree scope in writing — which subscriptions and management groups, which billing scope, the monthly spend band, who approves actions and who receives alerts — then take least-privilege access: Cost Management Reader for analysis, Cost Management Contributor at the scopes where budgets, alert rules and shared views have to be created, and Reader on subscriptions for resource-level review. Access is granted through GDAP you approve, consistent with our published access policy; no standing global admin, and no change rights until the guardrails and the approval path are agreed. We then build the twelve-month baseline, read Advisor's cost recommendations, inventory reservations, savings plans and Azure Hybrid Benefit claims, measure tag coverage, and confirm the cost-visibility policy where you are on CSP. The cycle closes with the onboarding note and a proposed guardrail set.

2. Guardrails live (first to second cycle)

Budgets go in per subscription, resource group or environment with actual and forecast thresholds and named recipients; anomaly alerts are created on every in-scope subscription; the tag standard is published with a coverage target and tag inheritance is switched on where the scope supports it; the showback view and the scheduled exports that feed it are built. We tune thresholds against the baseline so the first month's alerts are informative rather than noise, and we write down what each alert means and who acts on it. Where you want spend to be stopped rather than reported, we scope that automation separately and say plainly what Azure will and will not do on its own.

3. The monthly cost cycle

Each cycle follows the same shape: reconcile the month against budget and forecast and explain the variance by service, resource group and change; run the waste review and name the resources; triage every anomaly and threshold alert with a written cause; refresh the commitment and Azure Hybrid Benefit position; update tag coverage and the showback view; then produce the savings-actions list with owners, estimates, risks and change windows. Approved low-risk actions inside the agreed guardrails are executed by us; everything else is packaged for your change process and tracked. Questions between cycles go through IT Partner's intake with first response inside our published one-business-hour SLA; substantive analysis is scheduled into the cycle rather than improvised.

4. Commitment and renewal cycle

Coverage and utilization for reservations and savings plans are reviewed every cycle and in depth each quarter, against a stable-usage view rather than a single month's spike. Recommendations state the term, the scope, the payment shape and the exit consequences before you commit, because the exits differ: savings plans cannot be cancelled or exchanged, reservation refunds are capped by Microsoft at USD 50,000 of cancelled commitment in a rolling twelve-month window per billing scope, and from 1 February 2027 reservations purchased after that date for services covered by savings plans are no longer exchangeable (reservations bought before it keep one final exchange). Where you approve a purchase, we execute it through the appropriate path — Partner Center for customers on our CSP billing, or your own billing scope — against your written approval, and the reservation and Software Assurance calendar is updated.

5. Reporting, re-baseline and hand-off

The executive report goes out each cycle; each quarter we re-baseline the budgets, the commitment strategy, the unit-cost measures and the guardrails themselves, and review whether the tag standard still matches how the business is organised. Findings that need engineering are scoped into fixed-price written quotes you are free to place elsewhere. If you end the service, the artifacts stay where they were built — in your tenant, your storage and your reports — and the last cycle includes a hand-off note so whoever picks it up knows what each budget, alert rule and export is for.

Prerequisites

An Azure estate with steady, ongoing spend — typically in the region of $5,000 to $50,000 a month — on the CSP Azure plan under a Microsoft Customer Agreement, or on your own Microsoft Customer Agreement or Enterprise Agreement. A single virtual machine or a dormant subscription does not need a FinOps practice, and we will say so.
A named cost owner on your side — usually one finance and one IT lead — with an agreed approval path for the monthly savings-actions list and authority to decide when a workload owner disagrees with a finding.
Least-privilege access granted through GDAP or Azure RBAC that you approve: Cost Management Reader for analysis, Cost Management Contributor at the scopes where budgets and alert rules must be created (Cost Management Contributor or the equivalent permission is Microsoft's requirement for creating an anomaly alert rule), Reader on subscriptions for resource-level review, and a scoped role for executing approved changes if you want us to execute rather than hand off.
Agreement at onboarding on the tag standard, the coverage target, the alert recipients, the response window for anomalies, and the guardrails inside which we may execute approved low-risk actions without a further conversation.
Visibility of the commercial context that drives the numbers: existing reservations and savings plans and who bought them, Software Assurance and subscription-licence entitlements relevant to Azure Hybrid Benefit, any committed-spend agreement with Microsoft, and the renewal dates attached to each.
For customers on CSP: the cost-visibility policy enabled for your tenant so your own people can see Azure charges at pay-as-you-go rates in their own tenant — we enable it as part of onboarding where you want it, and it changes nothing about what you are invoiced.
The scope stated in your order: which subscriptions, and the monthly Azure spend band the fee covers. Estates above the agreed band, estates spanning multiple tenants, and estates with an unusually large number of subscriptions relative to their spend are quoted per estate rather than squeezed into a fixed fee.

Who does what

IT Partner

  • Run the cycle on schedule: baseline, guardrails, waste review, commitment and entitlement position, showback, savings-actions list, executive report — and say so plainly in the report when a cycle found little worth doing.
  • Configure and maintain budgets, cost alerts, anomaly alerts, exports and shared views in Microsoft Cost Management, and keep the tag standard, allocation model and action register current.
  • Triage anomaly and budget alerts inside the agreed response window and record a cause or an explicit 'expected spend' decision for each one.
  • Estimate each recommended action with the method stated, execute approved low-risk actions inside the agreed guardrails, and route everything else to your change process rather than around it.
  • Present commitment recommendations with the exit consequences stated — term, scope, payment shape, cancellation and exchange rules — and never purchase a reservation or savings plan without your written approval.
  • Report savings against your prior run rate with the method attached, publish no savings percentages of our own, and flag when a recommendation is uncertain rather than dressing it up.

Your team

  • Nominate the cost owner and approvers, and turn the monthly savings-actions list around inside your own cadence — an unapproved action saves nothing.
  • Own the tagging decisions: agree the tag set, name application and environment owners, and back the standard when a team would rather not use it.
  • Confirm licensing entitlements behind Azure Hybrid Benefit claims and any committed-spend obligations, and sign for reservation and savings-plan purchases.
  • Give workload owners the space to answer: several of the largest findings each month need a person who knows why a resource exists, and the practice stalls without them.
  • Route incidents, outages, patching and day-to-day administration through your support or managed-service arrangement rather than the monthly cost cycle — this service is not a response function.

What's not included

The one-time deep assessment — a full analysis of the estate with architecture-level opportunities, sized and prioritised, and workload-owner validation — is the Azure Cost Optimization and FinOps Assessment; the shorter fixed-price review of an existing environment is the Azure performance and cost optimization assessment. This service is the standing practice that follows either one, and does not repeat their depth in a monthly cycle.
Resource-health, performance, security and backup monitoring, patching and maintenance are Azure Resource Monitoring and Maintenance. That service watches whether the estate is healthy and includes cost observations in its reports; this one owns the cost practice — the allocation model, the guardrails, the commitment position, the executed actions and the number in front of finance. Where both run, neither duplicates the other's work and the reports are written to sit side by side.
Architecture rework and re-platforming: redesigning a workload, moving from IaaS to PaaS, database re-engineering, application refactoring, code and query optimization, and storage or network redesign. The cycle identifies and sizes these; delivery is a separate fixed-price project, and a foundation-level correction usually belongs with the Azure Landing Zone and Cloud Adoption Framework Implementation.
Azure consumption itself, reservation and savings-plan purchase amounts, marketplace charges and Microsoft licensing. These are Microsoft's charges, billed to you, and this fee never resells, discounts, caps or absorbs them. We recommend and, with your written approval, execute purchases through the appropriate path; the money and the commitment are yours.
Guaranteed savings, a percentage-of-savings or percentage-of-spend fee, a savings SLA, or any promise about the size of a reduction. This service is billed as a fixed monthly fee for the practice, and the savings figure in the report is a measurement rather than a target we have committed to hit.
Automated enforcement that stops spend: hard caps, automated shutdown or deallocation, quota-based blocking and policy-driven denial of expensive SKUs. Azure budgets notify rather than stop, and the automation behind an action group — or an Azure Policy set deployed as code — is a build we scope and quote separately.
Non-Azure spend: AWS, Google Cloud, other clouds, datacenter and colocation costs, telecom, and third-party SaaS subscriptions. The FinOps Foundation's 2025 Framework formally extends the practice to those scopes; this service is deliberately Azure-only. Microsoft 365 licensing waste is its own service, the Microsoft 365 License Optimization Subscription.
Licensing advice and entitlement determinations. We report which workloads claim Azure Hybrid Benefit and which look unsupported by an entitlement; confirming what your agreements actually permit, and any true-up or compliance position, is a licensing question for you or your licensing advisor — not a portal setting we can read.
Financial, tax and accounting advice, chargeback enforcement inside your organization, budget approval, and acting as your agent in a commercial negotiation with Microsoft. We produce the showback data and the recommendation; your finance function decides what to do with it and signs what needs signing.
Incident response, out-of-hours cover and a guaranteed response outside our published support hours. Cost alerts are triaged inside the agreed window during business hours; a runaway resource at 2am is an operational matter for your monitoring and support arrangement, not for the cost cycle.
Custom Power BI development, executive dashboards and data-warehouse work beyond the standard report and the scheduled exports. Where you want the data in your own store, a deployment of Microsoft's open-source FinOps toolkit (FinOps hubs and its Power BI reports) is a quoted build and its Azure resources are your consumption.
Third-party FinOps platform licences. The practice runs on Microsoft Cost Management, Azure Advisor and the exports your subscriptions already include; if you own a third-party cost platform we will work with it, but its licence and its administration are not part of this fee.

Limitations & technical notes

!Cost data is not real time, and no service can make it so. Azure cost and usage data is typically available within 8 to 24 hours; budgets are evaluated against it roughly every 24 hours and threshold emails normally arrive within an hour of that evaluation; cost anomaly detection compares each day's usage against a forecast built from the previous 60 days and runs about 36 hours after the end of the day (UTC) so that the data is complete. The practical consequence is stated plainly rather than buried: a resource that starts burning money on Monday morning is typically visible in an alert on Tuesday or Wednesday, not within minutes. Where minutes matter, the answer is an operational guardrail — policy, quota, automation — and we will tell you so.
!Budgets in Azure notify; they do not stop spend, and Azure offers no hard cap on a pay-as-you-go or CSP Azure-plan subscription. Anomaly detection is available at subscription scope, its sensitivity is not adjustable, an anomaly alert email is sent once at detection, alerts are sent based on the rule creator's access at the time of sending, and Microsoft does not offer cost anomaly alerts for Azure Government. Where any of these constraints matters to your risk, we design around it and say what the design costs.
!Savings are reported, never guaranteed. Estimates are made from utilization and rate data available at the time, with the method stated; actuals depend on your approvals, workload behaviour, Microsoft's prices and rate changes, existing discounts and whether a change survives the month. The report measures against your own prior run rate; we publish no return-on-investment or savings percentages of our own, and no figure in any report is a commitment about a future invoice.
!Commitments are the one part of this work that can cost you money, so the exits are stated before you buy rather than after. Savings plans cannot be cancelled or exchanged; reservations can be refunded, but Microsoft caps cancelled reservation commitment at USD 50,000 in a rolling twelve-month window per billing scope; and under Microsoft's published policy, reservations purchased on or after 1 February 2027 for services covered by savings plans are no longer exchangeable, while those purchased before that date retain one final exchange. Microsoft's reservation, savings-plan and pricing terms are Microsoft's and change without notice — we re-check them each cycle, and the version above was verified against Microsoft's documentation in September 2026.
!What you can see depends on your billing relationship, and it is worth understanding before the first report. On the CSP Azure plan, admin agents and billing admins in our partner tenant see costs at invoiced prices, while users in your own tenant see consumption computed at pay-as-you-go retail rates once the cost-visibility policy is enabled — in those customer-tenant views reservation usage shows as zero charge in both actual and amortized cost, so reservation economics have to be read at the billing scope. On your own Microsoft Customer Agreement or Enterprise Agreement you see your own negotiated position directly. Cost Management's built-in cost allocation rules for splitting shared costs are supported on Enterprise Agreement and Microsoft Customer Agreement scopes but not on Microsoft Partner Agreement scopes, so on CSP that split is produced in the report and the exported data instead.
!The practice depends on your organization answering questions. Tag coverage improves when application and environment owners are named and back the standard; right-sizing needs someone who knows why a machine is that size; and the largest single-line savings are usually decisions (turn this environment off, retire this system) that only your business can make. We chase, record and escalate; we do not overrule a workload owner, and a recommendation declined is recorded as declined rather than repeated silently every month.
!The fee covers the subscriptions and the monthly Azure spend band written into your order, billed per 30-day service period. Estates that grow past the agreed band, span multiple tenants, or carry an unusually large number of subscriptions relative to spend are re-quoted rather than absorbed — we would rather re-price openly than let the depth of the cycle quietly fall. Work that needs standalone analysis (a full assessment, a migration business case, a platform selection) is scoped separately and agreed in writing before it starts.
!This is a cost practice, not a support plan and not an operations function. Questions between cycles go through IT Partner's intake with first response within our published SLA of 1 business hour, and we publish our monthly support statistics for every month since December 2023. Organizations that buy their Microsoft licensing through IT Partner already have unlimited break-fix support during business hours under our published policy; this service neither replaces nor changes that, and 24/7 cover is a separately priced agreement.

Frequently asked questions

What is the Managed Azure FinOps and Cost Guardrails Service?

A recurring monthly practice that owns your Azure cost number. Each cycle we maintain the allocation model and tag standard, keep budgets and anomaly alerts configured and triaged in Microsoft Cost Management, run a waste review that names resources, track reservation and savings-plan coverage and utilization with recommendations before renewal dates, check Azure Hybrid Benefit and the entitlement behind it, maintain a showback and unit-cost view, produce a savings-actions list executed with your approval, and deliver an executive report stating what was actioned and what it changed. It costs $495 per month for the subscriptions and spend band agreed in your order.

Who is it for?

Finance and IT-operations leads at organizations with steady Azure spend — roughly $5,000 to $50,000 a month — where the bill is big enough to matter, moves in ways nobody can fully explain, and nobody's job description contains the words 'cloud cost'. It suits estates on the CSP Azure plan and estates on a customer's own Microsoft Customer Agreement or Enterprise Agreement equally. It does not suit a single server or a dormant subscription, and we will tell you that on the first call rather than sell you a cycle.

You already sell two Azure cost assessments. How is this different?

The assessments are one-time projects that tell you what is wrong and what it is worth: the Azure Cost Optimization and FinOps Assessment does the deep analysis with a sized savings roadmap, and the Azure performance and cost optimization assessment is the shorter fixed-price review of an existing environment. Both explicitly stop at recommendations. This service is what happens next, every month: the guardrails get built, the actions get executed or routed, the commitment position gets managed before renewal dates, and someone reports the number. Buying the assessment first is the normal sequence; buying this without one is fine too, since the first cycle builds its own baseline — it just goes narrower and deeper over more months rather than all at once.

How is this different from Azure Resource Monitoring and Maintenance, and can we run both?

Run both if you want both, and they will not overlap in practice. Azure Resource Monitoring and Maintenance watches whether the estate is healthy — resource health, performance, security posture, backup verification — and its reports include cost observations that come out of utilization data. This service owns the cost practice itself: the allocation model and tag standard, budgets and anomaly alerts as configured guardrails, the commitment and Azure Hybrid Benefit position, a monthly savings-actions list that actually gets executed, and an executive report written for finance. One tells you a VM is unhealthy; the other tells you it has been idle for 90 days, who owns it, what deleting it saves and when the change window is.

Do you guarantee savings? Would you work for a percentage of them?

No to both, on this service, deliberately. The fee is fixed at $495 a month whether the cycle finds a large saving or confirms the estate is already tight, which is exactly the incentive you want from the person telling you what to buy. A percentage-of-savings model rewards big, risky commitment purchases and quietly punishes the advice to do nothing this month. Savings are measured against your prior run rate with the method stated in the report; the figure that matters is the one on Microsoft's invoice, and it depends on your approvals, your workloads and Microsoft's prices as much as on our work.

What does $495 a month cover, and what happens if our estate is bigger?

It covers the full monthly cycle — baseline maintenance, allocation and tagging, guardrails and alert triage, the waste review, the commitment and Azure Hybrid Benefit position, showback and unit cost, the savings-actions list with approved execution inside the agreed guardrails, and the executive report — for the subscriptions and the monthly Azure spend band written into your order, billed per 30-day service period. Above that band, or where the estate spans multiple tenants or an unusually large number of subscriptions, we quote per estate rather than stretch a fixed fee and quietly thin the work. Azure consumption, reservations, savings plans and Microsoft licensing are billed separately by Microsoft and are never part of this fee.

Do we have to buy Azure through IT Partner for this to work?

No. The service works on your own Microsoft Customer Agreement or Enterprise Agreement with the roles you delegate, and on the CSP Azure plan whether or not we are the partner of record. Where your subscriptions are on the Azure plan under our CSP billing, we get some useful mechanics natively — costs at customer scope in our partner tenant at invoiced prices, budgets and reservation utilization alert rules created centrally with your people as recipients — but that is a convenience, not a requirement. If you do want billing in one place, switching your CSP billing is a no-charge, non-technical change that leaves your tenant and resources untouched.

Will our own team be able to see costs in our tenant?

That depends on a setting many CSP customers do not know exists, and checking it is part of onboarding. On the Azure plan, the cost-visibility policy that lets subscription users view Azure usage charges in their own tenant is disabled by default until the partner enables it; once enabled, anyone with the right Azure role can analyse costs, save views and set budgets for their own subscriptions and resource groups — computed at pay-as-you-go retail rates, which are the rates shown in Microsoft's public pricing calculator rather than what appears on an invoice. One quirk to know: in those customer-tenant views, reservation usage shows as zero charge in both actual and amortized cost, so reservation economics are read at the billing scope and reported to you by us.

How quickly will we know about a cost spike?

Usually a day or two, and we would rather say that than imply real time. Azure cost and usage data lands within roughly 8 to 24 hours; budgets are evaluated about every 24 hours; anomaly detection runs about 36 hours after the end of the day in UTC because it needs a complete day of data. So a runaway resource started on Monday typically surfaces in an alert on Tuesday or Wednesday, and we triage it inside the response window we agree with you. If your risk profile needs faster than that, the answer is not a better cost report — it is a guardrail in the platform: policy restrictions on expensive SKUs and regions, quotas, or automation that acts on the alert. We will scope that separately and tell you what it costs.

Can you put a hard cap on our Azure spend?

Not in the way most people mean, and no honest provider can. Azure budgets are notification instruments — they alert at actual and forecast thresholds, and they do not stop resources or block deployments. There is no hard spending cap on a pay-as-you-go or CSP Azure-plan subscription. What you can have is engineered: an action group behind a budget threshold that triggers a Logic App or Function to deallocate non-production resources, Azure Policy rules that deny expensive SKUs or regions, quotas that constrain a subscription, and non-production schedules that shut things down every night. Those are builds we scope and quote; this service designs them, recommends them and then operates the reporting around them.

Do you make changes in our environment, or only recommend?

Both, on your terms and inside guardrails we agree in writing at onboarding. Low-risk actions within those guardrails — deleting an unattached disk or an orphaned public IP, applying an agreed non-production schedule, adjusting a storage lifecycle rule — are executed by us once you approve the monthly list. Anything touching production capacity, a workload owner's environment, or anything your change process owns is packaged for that process and tracked until it closes. If you would rather we never touch the estate, that is a valid setup: we hand off the list and stay read-only, and you keep the execution.

Who buys reservations and savings plans, and who carries the risk?

You approve every purchase in writing and you carry the commitment — we never buy on your behalf without that approval, and nobody at IT Partner earns a commission on it. Where your subscriptions are on our CSP billing we can execute an approved purchase through Partner Center against your Azure plan; on your own Microsoft Customer Agreement or Enterprise Agreement you (or we, with delegated rights) purchase in your billing scope. What we owe you is the analysis before the signature: coverage against genuinely stable usage rather than a spike month, the term and scope that fit, the payment shape, and the exit consequences — because savings plans cannot be cancelled or exchanged and reservation refunds are capped by Microsoft at USD 50,000 of cancelled commitment in a rolling twelve-month window per billing scope.

Microsoft is changing reservation exchanges in 2027 — does that affect us?

It affects anyone planning to buy reservations, so it is part of every commitment recommendation we make. Under Microsoft's published policy, reservations purchased on or after 1 February 2027 are not eligible for exchange where the service is covered by savings plans — Azure Virtual Machines, App Service, Azure SQL Database and similar — while reservations purchased before that date retain the right to one final exchange. Reservations for products savings plans do not cover, such as Azure VMware Solution, are excluded from the change, instance size flexibility is unaffected, and the cancellation policy and its USD 50,000 rolling cap are unchanged. In practice it raises the price of guessing wrong on a three-year reservation and makes savings plans a better fit for workloads still evolving — which is a judgement we make with you, per workload, in writing. Microsoft's terms are Microsoft's; we re-check them each cycle.

What access do you need? Do you need global admin?

No global admin, and no standing rights. We ask for Cost Management Reader for analysis, Cost Management Contributor at the scopes where budgets, alert rules and shared views must be created (Microsoft requires Cost Management Contributor or the equivalent permission to create an anomaly alert rule), Reader on subscriptions for resource-level review, and — only if you want us to execute rather than hand off — a scoped role for the approved change types. Access is granted through GDAP that you approve, time-bound and least-privilege, against a published access policy you can compare any request against.

Our tagging is a mess. Is that a blocker?

It is the normal starting condition, and fixing it is the first month's work rather than a prerequisite. We agree a minimum tag set (usually owner, environment, and application or cost centre), measure current coverage honestly, switch on tag inheritance in Cost Management where your billing scope supports it so subscription and resource-group tags flow onto child resource usage records, and then chase the remaining untagged spend to a human month by month with the coverage figure in the report. Where you want the standard enforced rather than encouraged, Azure Policy rules do that — recommended here, deployed as a separate scoped project.

Do you cover AWS, Google Cloud or our SaaS spend?

No — this service is Azure-only, on purpose. The FinOps Foundation's 2025 Framework formally widened the discipline beyond public cloud into scopes such as SaaS, licensing, datacenter and AI, and we borrow its practice model, but a page that claims all of it is a page that does none of it well. Microsoft 365 licensing waste has its own recurring service, the Microsoft 365 License Optimization Subscription. For other clouds we will tell you plainly that we are not the right provider.

What tooling does this run on — do we need to license a FinOps platform?

No third-party platform and no agent. The practice runs on what your subscriptions already include: Microsoft Cost Management (cost analysis and smart views, budgets and cost alerts, anomaly detection, scheduled exports including the FOCUS dataset, reservation and price-sheet datasets) plus Azure Advisor's cost recommendations and the Azure resource data behind them. Where you want the data in your own store, Microsoft publishes an open-source FinOps toolkit — FinOps hubs and its Power BI reports — which we can deploy into your subscription as a quoted build; it is open-source tooling and the Azure resources it runs on are your consumption. If you already own a third-party cost platform, we will work inside it; its licence is yours.

How do you calculate what we saved?

Against your own prior run rate, with the method written next to the number. For a deletion or a de-allocation it is the resource's recent rated cost projected forward. For right-sizing it is the rate difference between the old and new SKU at observed usage. For a commitment it is the modelled difference between committed and on-demand rates for the covered usage, which is why we track utilization afterwards — an under-used reservation is a loss, not a saving. Estimates are shown before an action runs and actuals afterwards, and where a saving cannot be cleanly attributed (because the workload also changed) we say so instead of claiming it. No return-on-investment percentages of our own appear in any report.

How does billing work, and can we stop?

A flat $495 per month, invoiced per 30-day service period, with no long-term contract: stop any month and all we ask is payment of previously approved invoices. Nothing you keep is locked to us — the tag standard, budgets, alert rules, exports, showback model, action register and every report live in your tenant and your storage, and the final cycle includes a hand-off note. Projects that come out of the practice are quoted fixed-price in writing and paid after you approve delivery, and you are always free to place that work elsewhere.

Didn’t find your question?

Ask it here. A real engineer answers by email within one business day — and if it’s a good one, it becomes part of this page so the next person finds it.

Answered by a person, one time, to your inbox. Nothing you type here is published without a human reviewing and anonymizing it first.

Often combined with

$495 per month
30 days
Start monthly Azure FinOps