Need clarification regarding the recommendation logic, billing behavior, and validation process for Azure Compute Savings Plans and Azure App Service Reserved Instance recommendations.

Question

Need clarification regarding the recommendation logic, billing behavior, and validation process for Azure Compute Savings Plans and Azure App Service Reserved Instance recommendations.

RAKSHITHA BAI 0

Need clarification on the below queries:

How does Azure Advisor calculate the recommended hourly commitment for Azure Compute Savings Plans? Is it based on historical PAYG compute usage or discounted usage? Is there any official Microsoft documentation explaining the recommendation logic?
How is the Savings Plan billed? If the recommended hourly commitment is purchased, is the committed amount billed every month, or is the final bill based on discounted compute consumption? How is the estimated savings reflected in the invoice? Please share any official billing example or documentation.
A purchased Savings Plan shows 100% utilization. Is there a Microsoft-supported way to identify which Azure resources (VMs, AKS, App Service Plans, etc.) are consuming the Savings Plan benefit? Is there any portal view or report that shows the matched resources and the applied Savings Plan benefit?
Can the Savings Plan recommendation be validated using Azure Cost Analysis? Which filters should be used, and should the compute cost in Cost Analysis match the recommended monthly commitment?
Azure Advisor recommends 9 App Service Reserved Instances, while more App Service Plans exist in the subscription. How is the recommended reservation quantity calculated? Is there any official Microsoft documentation explaining this recommendation logic?

0 comments

3 answers

Your answer

Answer 1

Hi, treat recommendations as usage-based estimates, not purchase guarantees. Savings Plan recommendations analyze eligible compute spend over a lookback window and propose an hourly commitment; reservation recommendations use eligible SKU/region usage and scope. App Service reservations apply to supported App Service stamp/plan meter usage, not every charge shown under the web app. Validate by exporting cost details, grouping by meter/resource/region, and modeling the proposed commitment against a representative busy and quiet period. Purchases are billed for the commitment whether or not you fully use it, while benefit allocation follows the selected scope. Before buying, confirm the exact eligible meters and scope in Cost Management and have Billing support review any recommendation that cannot be reconciled to your usage.

Answer 2

hi RAKSHITHA BAI & thx for sharing urs issue here at Q&A portal,

soooo... a lit bit of theory : ) Azure Advisor bases the Savings Plan recommendation on recent hourly PAYG usage and cost for eligible compute resources, normally using 7, 30, or 60-day lookback periods. Azure simulates different hourly commitments and highlights the one expected to produce the highest savings for the selected term and scope. It isn’t calculated from already discounted Savings Plan usage. A Savings Plan is a fixed hourly commitment for one or three years. That committed amount is charged whether u fully use it or not. Eligible compute usage consumes the benefit each hour at discounted rates, usage above the commitment is billed at PAYG rates, and unused hourly commitment doesn’t roll over.

To identify which resources consumed the benefit, use Cost Management with the Amortized cost view. That allocates the Savings Plan cost to the resources that received the benefit and can expose unused commitment separately. The utilization percentage only tells u how much of the commitment was consumed, not by itself which resources consumed it.

Cost Analysis can help validate the recommendation, but the total compute cost won’t necessarily equal the recommended monthly commitment. The recommendation is based on the hourly usage shape, not just monthly spend. Two subscriptions can spend the same amount monthly but have very different safe hourly commitments if one has steady usage and the other has spikes. For App Service reservations, Advisor looks at historical hourly usage and simulates reservation quantities to find the quantity expected to maximize savings. That’s why it may recommend 9 instances even though more App Service Plans exist. Only matching eligible instances, SKUs, regions, scopes, and sustained usage count toward the recommendation.

I collect some docs for u, i hope it would be helpfull https://learn.microsoft.com/azure/cost-management-billing/savings-plan/purchase-recommendations & https://learn.microsoft.com/azure/cost-management-billing/savings-plan/choose-commitment-amount & https://learn.microsoft.com/azure/cost-management-billing/savings-plan/savings-plan-overview & https://learn.microsoft.com/azure/cost-management-billing/reservations/reserved-instance-purchase-recommendations & https://learn.microsoft.com/azure/cost-management-billing/reservations/prepay-app-service

rgds,

Alex

&

If my answer was helpful pls mark it and additional thx if u follow me at Q&A portal

and at my blog https://ctrlaltdel.blog/

Answer 3

Suchitra Suregaunkar 16,125 Microsoft External Staff Moderator

Hello RAKSHITHA BAI

Thank you for reaching out and for the detailed questions. Happy to clarify each of them below with the official Microsoft references.

How Azure Advisor calculates the recommended hourly commitment for Compute Savings Plans:

The recommendation is based on your actual on-demand (pay-as-you-go) hourly usage and cost from Savings Plan–eligible compute resources, not discounted usage. Advisor analyses your hourly PAYG cost over the last 30 days (the API also supports 7 and 60 days), then runs simulations to identify the hourly commitment that produces the maximum net savings. It also runs a shorter 3-day simulation and picks the lower of the two recommendations to prevent overcommitment from stale data. Existing reservations and savings plans are excluded from the calculation.

Reference:

How savings plan recommendations are generated

Choose a savings plan commitment amount

How the Savings Plan is billed and how savings appear on the invoice

A Savings Plan is a billing offer you commit to a fixed hourly amount for 1 or 3 years and get discounted rates on eligible compute. You are billed for the full hourly commitment for every hour of the term, whether you use it or not. Eligible usage within the commitment is charged at the discounted Savings Plan rate; any usage above the hourly commitment is billed at PAYG rates. Unused commitment in an hour does not roll over.

On the invoice:

Purchase charges appear as ChargeType = Purchase and PricingModel = SavingsPlan.
In Actual Cost data, the EffectivePrice for usage covered by the Savings Plan is zero (already paid via the hourly commitment).
In Amortized Cost data, the EffectivePrice shows the prorated Savings Plan cost attributed per resource, so you can see the monetary value each resource consumed.
Payment can be upfront or in equal monthly instalments at no extra cost.

Savings are realised as avoided cost , no separate "discount" line item appears, the higher PAYG line is simply replaced with the discounted (or zero) line.

Reference:

What are Azure savings plans?

View savings plan cost and usage details

Calculate EA Savings Plan savings — worked example

Identifying which resources are consuming the Savings Plan benefit (100% utilization)

Utilization % alone doesn't show which resources are consuming the benefit.

To identify the exact matched resources (VMs, AKS nodes, App Service Plans, Container Apps, Functions Premium, etc.), please use Cost Analysis with Amortized Cost:

Go to Cost Management + Billing → Cost analysis.
Change the metric from Actual cost → Amortized cost.
Add filter: Pricing model = SavingsPlan.
Set Group by = Resource (or Subscription / Resource group).
Set Granularity = Daily or Monthly and chart type = Table.

Each row shows the resource consuming the benefit and the prorated cost. The BenefitId / BenefitName fields identify the specific Savings Plan.

For historical utilization %, use Savings plans → Utilization in the Azure portal.

Charge back savings plan costs

View savings plan utilization

Validating the Savings Plan recommendation using Cost Analysis.

Yes, Cost Analysis is the right tool to validate. However, please note the compute cost in Cost Analysis will not exactly equal the recommended monthly commitment , Advisor deliberately recommends a commitment that covers your steady-state hourly usage, not your peak, so you don't over-commit during off-peak hours.

Suggested filters to validate:

Setting	Value
Metric	Actual cost (to see raw PAYG being targeted)
Filter — Pricing model	On demand
Filter — Service name	Virtual Machines, App Service, Functions, Container Instances, Container Apps, Dedicated Host, Spring Apps
Granularity	Daily
Group by	Service / Resource

Divide the daily PAYG compute cost by 24 to get the approximate hourly PAYG spend, then compare it against Advisor's recommended hourly commitment. Advisor's value should sit at or below your consistent hourly baseline.

Why Advisor recommends only 9 App Service Reserved Instances even though more App Service Plans exist

The recommended quantity is not based on the total count of App Service Plans in the subscription. It's calculated from actual hourly usage of eligible instances over the past 7, 30, and 60 days. The engine simulates cost with and without reservations at different quantities and recommends the quantity that maximizes savings. Instances are excluded if they:

Are on non-eligible tiers (only Premium v3 and above and Isolated v2 and above are eligible for App Service RIs)
Run intermittently, are stopped/deallocated, or autoscale down
Are already covered by an existing reservation or savings plan

So a recommendation of 9 typically means Advisor detected 9 eligible App Service Plan instances running consistently at eligible SKUs during the look-back window. Other App Service Plans in the subscription are likely on lower tiers, running intermittently, or already covered.

Reservation recommendations — calculation logic

Save costs with Azure App Service reserved instances

Hope this clarifies each of your questions. Please feel free to reach out if any point needs further explanation, happy to help.

If you found the comment helpful, please consider clicking "Upvote it".

Thanks,

Suchitra.

RAKSHITHA BAI 0 Reputation points

2026-07-26T15:45:55.2766667+00:00
Thank you for the detailed explanation and the official documentation references.

Based on the information shared, we have understood the recommendation logic, billing behavior, and the difference between Actual Cost and Amortized Cost.

However, we have a few additional clarifications:

1. Savings Plan realized savings

We understand that Amortized Cost with Pricing Model = SavingsPlan helps identify the resources consuming the Savings Plan benefit.

Is there any Microsoft-supported method or report available to determine the actual savings realized from an existing Savings Plan without manually comparing the PAYG and effective prices from the Cost Details export?

2. Azure Advisor recommendation validation

While validating the Advisor recommendation, apart from the filters below:

Metric: Actual Cost

Pricing Model: OnDemand

Service Family: Compute

Granularity: Daily

Group By: Resource/Service

Are there any additional Cost Analysis filters or meter-level filters recommended by Microsoft to accurately identify the eligible on-demand compute usage that contributes to the Savings Plan recommendation?

3. Calculating realized Savings Plan savings

Is the Cost and usage details (Actual) and Cost and usage details (Amortized) export the recommended approach to calculate the realized Savings Plan savings?

If so, could you please share the relevant fields (for example, EffectivePrice, UnitPrice, BenefitId, BenefitName, PricingModel, etc.) that should be compared to derive the exact savings?

4. Azure Advisor recommendation with existing Savings Plans

The tenant currently has two active Azure Savings Plans.

We observed that one of the Savings Plans has a Last 7-day utilization of 96%, indicating that it is not being fully utilized at all times. Despite this, Azure Advisor is recommending the purchase of an additional Savings Plan.

Could you please help us understand the recommendation logic in this scenario? Specifically:

Why is an additional Savings Plan being recommended when one of the existing Savings Plans is not consistently achieving 100% utilization?

Does Advisor evaluate uncovered eligible PAYG compute usage independently of the utilization of existing Savings Plans?

Under what conditions would Advisor still recommend an additional Savings Plan even when an existing plan shows less than 100% utilization?
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.

Suchitra Suregaunkar 16,125 Microsoft External Staff Moderator

Hello RAKSHITHA BAI

Thanks for confirming, and glad the earlier explanation on recommendation logic and Actual vs. Amortized cost was useful. Let me take your remaining points one by one.

Determining actual realized savings without manually comparing prices:

I want to be upfront here so there's no confusion later: Azure does not currently provide a built-in portal report that shows realized monetary savings for a Savings Plan. What the portal gives you natively is utilization percentage, not a savings figure.

The Microsoft-supported method to determine realized savings is to derive it from the amortized cost data. This isn't a workaround, Microsoft publishes a dedicated step-by-step article for exactly this calculation:

Calculate EA savings plan cost savings

If you'd rather automate it instead of working in Excel each month, these are the supported programmatic options:

Cost Details / Usage Details API with metric=AmortizedCost — returns the same fields for scripted calculation. [docs.azure.cn]

Benefit Utilization Summaries API — for utilization trend data. [learn.microsoft.com]

Scheduled Exports (Cost Management → Exports → Amortized cost) — pushes the dataset to storage daily so you can build a recurring Power BI report. [docs.azure.cn]

So: utilization is out-of-the-box, realized savings must be derived. That is the current design.

Additional Cost Analysis and meter-level filters:

Your existing filter set is a reasonable starting point, but Service Family = Compute is broader than Savings Plan eligibility, so it will overstate the eligible baseline and make the Advisor recommendation look inconsistent with your numbers. Two refinements will fix that.

a) Narrow to Savings Plan–eligible services. Savings plan for compute covers infrastructure costs for: Azure Virtual Machines, Azure App Service, Azure Functions (Premium plan), Azure Container Instances, Azure Dedicated Host, Azure Container Apps, and Azure Spring Apps for Enterprise.

b) Exclude non-eligible meters. Savings plan does not cover software, networking, or storage charges. This matters a lot in practice, because Windows and SQL licensing meters sit right alongside VM compute meters and will inflate your baseline if left in.

Suggested refined filter set:

Filter	Value
Metric	Actual Cost
Pricing Model	OnDemand
Service Name	VMs, App Service, Functions, Container Instances, Container Apps, Dedicated Host, Spring Apps
Meter / Meter Category	Exclude Windows & SQL licensing, managed disks/storage, bandwidth/networking
Granularity	Daily
Group by	Service Name, then Meter

One important point: Cost Analysis is not the authoritative source for eligibility. The definitive list of eligible products and meters is in your Savings Plan price sheet, and Microsoft directs customers to download it to confirm product inclusion:

Troubleshoot savings plan utilization — benefit scope & product inclusion

What are savings plans? — covered services and exclusions

Is the Cost and usage details export the recommended approach, and which fields to compare?

Yes — the Cost and usage details (Amortized) export is the recommended dataset.

Here is the procedure and the fields involved.

Step 1 — Download the file Cost Management + Billing → Usage + charges → select the month → Download → under Usage Details choose Amortized charges (usage and purchases) → Prepare document → Download CSV.

Step 2 — Filter the data

benefitName → clear Blanks (keeps only records that received a Savings Plan benefit)

ChargeType → select Usage only (removes purchase rows)

Step 3 — Calculate the savings Insert a column next to UnitPrice and apply:

TotalUsedSavings = (UnitPrice − EffectivePrice) × Quantity

Sum that column. The total is your realized savings for the period.

Field reference:

Field	What it represents
`UnitPrice`	Your on-demand (PAYG) rate, including any negotiated discounts
`EffectivePrice`	Actual dataset = `0` for SP-covered usage; Amortized dataset = prorated hourly SP cost
`Quantity`	Units of usage consumed
`BenefitId` / `BenefitName`	The Savings Plan ID and name that delivered the benefit
`PricingModel`	`SavingsPlan` on SP-related purchase and usage records
`ProductOrderId` / `ProductOrderName`	Savings plan order ID and product name
`Term`	1-year or 3-year
`ChargeType`	`Purchase` (Actual dataset only) vs.`Usage`

Two caveats worth noting before you start:

The Amortized dataset contains no purchase records. If you need to reconcile the commitment charge itself against the invoice, use the Actual dataset filtered on ChargeType = Purchase.

If your billing account is MCA rather than EA, the amortized file does not includeUnitPrice. You'll need to pull on-demand unit prices from your MCA price sheet and join them before running the formula.

View savings plan cost and usage details

Charge back savings plan costs

Why Advisor recommends another plan while an existing plan sits at 96% utilization: This is expected behaviour, and it makes sense once you look at how the benefit is applied hourly rather than daily.

Why the recommendation appears despite <100% utilization:

Savings Plan benefit is applied hourly and is use-it-or-lose-it — each of the 24 hours in a day is a separate benefit window. Microsoft documents this exact scenario: you can incur on-demand charges while your savings plan utilization is below 100%, because in some hours the commitment goes partly unused, while in other hours your eligible usage exceeds the commitment and spills over to PAYG rates.

So 96% over 7 days does not mean you have 4% spare headroom available across all workloads. It means a small amount of commitment expired unused in specific low-usage hours and separately, other hours very likely had meaningful uncovered PAYG compute. It is that uncovered PAYG usage that is driving the new recommendation.

Does Advisor evaluate uncovered PAYG usage independently?

Yes. Recommendations are generated from your actual on-demand usage and costs on Savings Plan–eligible resources. Usage already covered by an existing plan is no longer on-demand usage, so it never enters the calculation. Advisor also reports what percentage of your compute cost is already covered by existing reservations and savings plans alongside each recommendation.

Microsoft states this directly for the reservation engine as well:"The recommendations account for existing reservations and savings plans. So, previously purchased reservations and savings plans are excluded when providing recommendations."

Conditions where Advisor still recommends moreL

Condition	Why it triggers a recommendation
Residual uncovered PAYG usage	Simulations found a commitment that yields net savings. Only net-savings candidates are ever surfaced. [learn.microsoft.com]
Narrow benefit scope	If a plan is scoped to a single subscription instead of Shared, eligible usage elsewhere is entirely uncovered. [learn.microsoft.com]
Reservations applied first	Azure applies reservation benefits before savings plan benefits, which can leave a plan partly unused while other usage stays on PAYG. [learn.microsoft.com]
Recent purchase not yet reflected	Allow at least 7 days after any SP or reservation purchase for recommendations to fully update. [learn.microsoft.com]

Wait 7 days after your most recent purchase, then re-check the recommendation.

Check the scope of both existing plans. If either is subscription-scoped, changing it to Shared often lifts utilization at no additional spend, worth trying before committing more.

Review the hourly utilization chart, not the 7-day average, to see the real coverage pattern.

If you do proceed, purchase at or below the recommended amount. Microsoft specifically warns that committing above the recommended value leads to underutilization and undermines the savings goal.

Thank You.