Oak & Mesa / FinOps

Azure FinOps Audit

Prepared for Lumen Trail Software LLC (EXAMPLE DATA)
Prepared by Zachery Kuykendall
12:54 PM CT, November 13, 2026
Period Aug 1, 2026 – Oct 31, 2026
EXAMPLE DATA — fictional company
EXAMPLE DATA. This report was generated from a synthetic Azure Cost Management export for a fictional company (Lumen Trail Software LLC (EXAMPLE DATA)). It is a demonstration, not a real client engagement, and the dollars are invented.

A read-only review of the Azure Cost Management export for Aug 1, 2026 – Oct 31, 2026 (92 days). Total spend in the period is $13,083.04. Rule-based findings below identify likely waste and rate-optimization opportunities. Every savings figure is an ESTIMATE with its assumption stated. Nothing in this report changes the client’s Azure environment.

Period spend
$13,083.04
Aug 1, 2026 – Oct 31, 2026 · 92 days
Last complete month
$5,467.47
October 2026 · prior $3,791.10
Avg daily
$142.21
5 anomalies flagged
Est. monthly savings
$705.91
Estimate conservative, non-overlapping stack

Executive summary

10 findings were generated from the export. The conservative, non-overlapping monthly savings total is $705.91 Estimate. That total includes Savings Plan estimates at a conservative 20% (inside Microsoft’s published 11%–65% compute Savings Plan range), dev/test shutdown estimates, and the current run-rate of resources that look idle — each of which is a maximum until attachment, association, or backup need is confirmed. A Reservation alternative on the same steady VMs is shown in the finding and is not added on top of the Savings Plan figure. Spike and tagging findings have no dollar claim.

ID Finding Severity Monthly savings
F-COMMIT-SP 4 pay-as-you-go compute resource(s) look like Savings Plan / Reservation candidates High $322.40 ESTIMATE
F-DEVTEST-247 2 dev/test compute resource(s) appear to run 24×7 High $199.28 ESTIMATE
F-ANOMALY 5 anomalous spend day(s) in the period High —
F-EGRESS-SPIKE Egress / bandwidth spikes detected High —
F-LOG-SPIKE Log Analytics / Azure Monitor spikes detected High —
F-DISK-IDLE 2 managed disk(s) look idle (capacity cost, essentially no transactions) Medium $68.20 ESTIMATE
F-PREMIUM-SKU 1 Premium storage resource(s) may be candidates for a lower SKU Medium $61.47 ESTIMATE
F-SNAP-OLD 2 snapshot(s) appear at least 30 days old (lower bound from the export) Medium $43.40 ESTIMATE
F-PIP-REVIEW 3 public IP address(es) are incurring charges — association CAN'T VERIFY Medium $11.16 ESTIMATE
F-UNTAGGED Untagged spend is 5% of the period ($663.28 USD) Low —
This is a findings report from a billing export, not a commitment to save a specific amount and not a Microsoft quote. Discounts vary by SKU, region, and term. Idle-resource savings require a portal or Resource Graph check before anything is deleted. The only input used was the client’s Cost Management CSV — no Azure credentials were required or used.

Spend summary and trend

Daily spend over the export window. Month-over-month figures use complete calendar months when the file covers them; otherwise a note is shown.

$0.00$70.10$140.21$210.31$280.41 Aug 1Sep 1Oct 1Oct 31
Month Spend Change
Aug 2026 $3,824.47 —
Sep 2026 $3,791.10 $-33.37 (-1%)
Oct 2026 $5,467.47 $+1,676.37 (+44%)

Pricing model mix

ModelSpendShare
OnDemand $12,531.04 96%
Reservation $552.00 4%

Untagged spend

5% of period spend ($663.28) has no resource tags. This is a governance finding, not a direct savings claim.

Charge type

TypeSpend
Usage$13,083.04

Top services, resource groups, and resources

Top services (MeterCategory)

Virtual Machines$4,416Log Analytics$2,625Azure App Service$1,840Storage$1,569SQL Database$1,380Bandwidth$910.00Application Gateway$310.00Networking$33.12

Top resource groups

rg-prod-compute$4,784rg-prod-data$4,080rg-shared-logs$2,625rg-dev-sandbox$552.00rg-test-qa$368.00rg-leftovers$353.28rg-prod-network$321.04

Top resources by period spend

Resource Resource group Service Spend Share
law-shared-01 rg-shared-logs Log Analytics $2,625.00 20.1%
app-prod-web-01 rg-prod-compute Azure App Service $1,840.00 14.1%
sqldb-prod-01 rg-prod-data SQL Database $1,380.00 10.5%
vm-prod-api-01 rg-prod-compute Virtual Machines $1,104.00 8.4%
vm-prod-api-02 rg-prod-compute Virtual Machines $1,104.00 8.4%
stprodassets rg-prod-data Bandwidth, Storage $1,048.00 8.0%
vm-prod-app-01 rg-prod-compute Virtual Machines $736.00 5.6%
disk-overprovisioned-p40 rg-prod-data Storage $608.12 4.6%
vm-dev-box-01 rg-dev-sandbox Virtual Machines $552.00 4.2%
vm-prod-sql-01 rg-prod-data Virtual Machines $552.00 4.2%
disk-prod-data-p30 rg-prod-data Storage $487.60 3.7%
vm-qa-worker-01 rg-test-qa Virtual Machines $368.00 2.8%
agw-prod-01 rg-prod-network Application Gateway $310.00 2.4%
disk-orphaned-std-01 rg-leftovers Storage $124.20 0.9%
snap-old-data-jan rg-leftovers Storage $82.80 0.6%

Month-over-month movers

Comparing Sep 2026 ($3,791.10) to Oct 2026 ($5,467.47). Basis: complete_months.

Services — largest increases

ServicePriorCurrentΔ
Log Analytics $540.00 $1,620.00 $+1,080.00 (+200%)
Application Gateway $0.00 $310.00 $+310.00
Bandwidth $240.00 $422.00 $+182.00 (+76%)
Virtual Machines $1,440.00 $1,488.00 $+48.00 (+3%)
Storage $510.30 $531.31 $+21.01 (+4%)
Azure App Service $600.00 $620.00 $+20.00 (+3%)
SQL Database $450.00 $465.00 $+15.00 (+3%)
Networking $10.80 $11.16 $+0.36 (+3%)

Services — largest decreases

No material decreases.

Resource groups — largest increases

Resource groupPriorCurrentΔ
rg-shared-logs $540.00 $1,620.00 $+1,080.00
rg-prod-network $3.60 $313.72 $+310.12
rg-prod-data $1,272.30 $1,492.71 $+220.41
rg-prod-compute $1,560.00 $1,612.00 $+52.00
rg-dev-sandbox $180.00 $186.00 $+6.00
rg-test-qa $120.00 $124.00 $+4.00
rg-leftovers $115.20 $119.04 $+3.84

Findings

Each dollar figure is labeled Estimate. The assumption that produced it is written under the finding. Figures included in the executive total are a non-overlapping stack (dev/test shutdown is not also counted as a Savings Plan; Reservation is an alternative, not an add-on).

F-COMMIT-SP · Rate optimization · High

4 pay-as-you-go compute resource(s) look like Savings Plan / Reservation candidates

$322.40
ESTIMATE / month

Recommendation. Start with a 1-year Azure savings plan for compute on this steady OnDemand spend (flexibility across SKUs and regions). The conservative ESTIMATE in the total is 322.40 USD/month at 20%. A Reservation on a stable SKU could save more — ESTIMATE 580.32 USD/month at the 36% published low end — but only if the exact size, family, and region will not change. Confirm with Azure Advisor recommendations before buying. Sources: https://azure.microsoft.com/en-us/pricing/offers/savings-plans and https://azure.microsoft.com/en-us/pricing/offers/reservations.

Assumption / basis. These VMs (or compute-hour resources) show steady ≥20 hours/day on at least 85% of the days in their span, for at least 21 days, billed OnDemand (pay-as-you-go). Savings Plan ESTIMATE uses a conservative 20% discount, inside Microsoft's published compute Savings Plan range of 11%–65%. Microsoft publishes a compute Savings Plan range of 11%–65% vs pay-as-you-go (65% is a 3-year term on one M64dsv2 Ubuntu VM in East US; Azure pricing as of January 2026). Source: azure.microsoft.com/en-us/pricing/offers/savings-plans Reservation ESTIMATE uses 36%, the low end of Microsoft's published Reserved VM Instance range of 36%–72%. Microsoft publishes a Reserved VM Instance range of 36%–72% vs pay-as-you-go (72% is a 3-year term on one M64ls SUSE Linux VM in East US; Azure pricing as of January 2026). Source: azure.microsoft.com/en-us/pricing/offers/reservations Actual discounts vary by SKU, region, and 1- vs 3-year term. Dev/test-named resources are excluded here and handled under a shutdown finding. Monthly figures use the October 2026 run-rate (a complete calendar month in the export).

Confidence. ESTIMATE. Usage pattern is observed from the export; the discount % is a conservative assumption inside Microsoft's published ranges, not a quote. Alternative Reservation ESTIMATE (not added on top): 580.32 USD/month.

resource resource group subscription median daily hours steady days span days monthly ondemand cost sp estimate ri estimate
app-prod-web-01 rg-prod-compute EXAMPLE DATA - Lumen Trail Production (FICTIONAL) 24.00 92 92 $620.00 $124.00 $223.20
vm-prod-api-01 rg-prod-compute EXAMPLE DATA - Lumen Trail Production (FICTIONAL) 24.00 92 92 $372.00 $74.40 $133.92
vm-prod-api-02 rg-prod-compute EXAMPLE DATA - Lumen Trail Production (FICTIONAL) 24.00 92 92 $372.00 $74.40 $133.92
vm-prod-app-01 rg-prod-compute EXAMPLE DATA - Lumen Trail Production (FICTIONAL) 24.00 92 92 $248.00 $49.60 $89.28
F-DEVTEST-247 · Scheduling · High

2 dev/test compute resource(s) appear to run 24×7

$199.28
ESTIMATE / month

Recommendation. Put auto-shutdown or a start/stop schedule on these dev/test VMs (Azure DevTest Labs, Automation, or Native auto-shutdown). Confirm with the owners before changing hours. If a box truly must run 24×7, consider a Savings Plan instead.

Assumption / basis. Resources were classified as non-production because the resource group, resource name, subscription, or an env/environment tag matched dev, test, qa, uat, sandbox, staging, or similar. They also meet the 24×7 usage test (≥20 hours/day on ≥85% of days). Savings ESTIMATE assumes a weekday 12-hour schedule (60 hours on / 168 hours in a week), i.e. a 64.3% reduction in compute hours, and that nights and weekends do not need coverage. This is an assumption, not a Microsoft-published figure. Monthly figures use the October 2026 run-rate (a complete calendar month in the export). These resources are excluded from the Savings Plan total so the two findings are not double-counted.

Confidence. ESTIMATE, subject to whether a 12-hour weekday schedule is acceptable.

resource resource group subscription median daily hours monthly compute cost shutdown estimate
vm-dev-box-01 rg-dev-sandbox EXAMPLE DATA - Lumen Trail Development (FICTIONAL) 24.00 $186.00 $119.57
vm-qa-worker-01 rg-test-qa EXAMPLE DATA - Lumen Trail Development (FICTIONAL) 24.00 $124.00 $79.71
F-ANOMALY · Anomaly · High

5 anomalous spend day(s) in the period

No dollar claim

Recommendation. Inspect the driving meter and resource on each day (listed below). If a spike is a reservation purchase, it may be expected on an ActualCost export. If it is usage, look for a deployment, a data job, or a runaway diagnostic setting.

Assumption / basis. A day is flagged when total spend is ≥ 2.5 standard deviations above the period mean, or at least 1.5× the trailing 7-day median, and above the mean. No recurring monthly savings are claimed; some spikes are one-time purchases or incidents.

Confidence. Observed from daily totals in the export.

date cost zscore vs trailing median top driver
2026-10-18 $250.37 3.63 1.53 stprodassets (Bandwidth, $95.00)
2026-10-19 $250.37 3.63 1.05 stprodassets (Bandwidth, $95.00)
2026-10-12 $238.37 3.23 1.46 law-shared-01 (Log Analytics, $120.00)
2026-10-13 $238.37 3.23 1.46 law-shared-01 (Log Analytics, $120.00)
2026-10-14 $238.37 3.23 1.46 law-shared-01 (Log Analytics, $120.00)
F-EGRESS-SPIKE · Anomaly / networking · High

Egress / bandwidth spikes detected

No dollar claim

Recommendation. Identify what sent traffic on the spike days (storage egress, a public app, a cross-region backup, or a data-transfer job). Consider Private Link, region alignment, and caching if this is becoming a baseline. Recent monthly run-rate is $422.00 USD.

Assumption / basis. No recurring monthly savings are claimed — a spike may be a one-time event or a new baseline. Days are flagged with the same anomaly rule as the all-spend anomaly section (z-score or 1.5× the trailing 7-day median). Savings from lowering Log Analytics retention or reducing egress CAN'T VERIFY without the workspace retention setting and traffic owners. October 2026 spend was $422.00 vs $240.00 in September (+$182.00, 76%).

Confidence. Observed from the export. Dollar savings CAN'T VERIFY without configuration context. October 2026 spend was $422.00 vs $240.00 in September (+$182.00, 76%).

date cost zscore
2026-10-18 $95.00 6.71
2026-10-19 $95.00 6.71
F-LOG-SPIKE · Anomaly / telemetry · High

Log Analytics / Azure Monitor spikes detected

No dollar claim

Recommendation. Review workspace data sources, diagnostic settings, and retention (90 days is a common default; dropping interactive retention often cuts cost). Investigate the spike days for a noisy resource or a new diagnostic setting. Recent monthly run-rate is $1,620.00 USD.

Assumption / basis. No recurring monthly savings are claimed — a spike may be a one-time event or a new baseline. Days are flagged with the same anomaly rule as the all-spend anomaly section (z-score or 1.5× the trailing 7-day median). Savings from lowering Log Analytics retention or reducing egress CAN'T VERIFY without the workspace retention setting and traffic owners. October 2026 spend was $1,620.00 vs $540.00 in September (+$1,080.00, 200%).

Confidence. Observed from the export. Dollar savings CAN'T VERIFY without configuration context. October 2026 spend was $1,620.00 vs $540.00 in September (+$1,080.00, 200%).

date cost zscore
2026-10-12 $120.00 4.30
2026-10-13 $120.00 4.30
2026-10-14 $120.00 4.30
2026-10-01 $45.00 0.77
2026-10-02 $45.00 0.77
2026-10-03 $45.00 0.77
2026-10-04 $45.00 0.77
F-DISK-IDLE · Idle / orphaned resources · Medium

2 managed disk(s) look idle (capacity cost, essentially no transactions)

$68.20
ESTIMATE / month

Recommendation. In the Azure portal or Resource Graph, confirm each disk is Unattached (or attached to a stopped-deallocated VM you no longer need). Snapshot if required, then delete. Do not delete a disk that is actually in use.

Assumption / basis. ESTIMATE equals the current monthly run-rate, which is the *maximum* savings if the disk is unattached (or otherwise unneeded) and is deleted after a snapshot is kept only if required. Attachment status CAN'T VERIFY from a Cost Management export — managed disks have their own ResourceId and incur capacity charges whether or not they are attached. The heuristic is: disk-like meters with capacity cost over the window and ~zero Operations/Transactions/IOPS meters. Confirm in the portal or with Azure Resource Graph (`disk.diskState == Unattached`) before deleting. Monthly figures use the October 2026 run-rate (a complete calendar month in the export).

Confidence. CAN'T VERIFY attachment from the CSV. The dollar figure is an ESTIMATE of maximum monthly savings if deletion is appropriate.

resource resource group subscription monthly cost transaction quantity days present
disk-orphaned-std-01 rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $41.85 0.00 92
disk-orphaned-std-02 rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $26.35 0.00 92
F-PREMIUM-SKU · SKU right-size · Medium

1 Premium storage resource(s) may be candidates for a lower SKU

$61.47
ESTIMATE / month

Recommendation. For each flagged disk, check Metrics (IOPS, throughput, latency) and the attached VM's requirements. If they fit Standard SSD or a smaller Premium size, price the change in the Azure Pricing Calculator before converting.

Assumption / basis. Illustrative ESTIMATE of a 30% reduction if a Premium SSD (or similar premium tier) with low observed transaction/IOPS meters is moved to Standard SSD (or a smaller Premium size). 30% is NOT a Microsoft-published average — exact SKU prices vary by size, tier, and region. The exact savings CAN'T VERIFY from the export alone; confirm IOPS/throughput needs and price the change in the Azure Pricing Calculator. Monthly figures use the October 2026 run-rate (a complete calendar month in the export).

Confidence. ESTIMATE with a stated illustrative percentage; exact SKU delta CAN'T VERIFY without retail prices and performance requirements.

resource resource group monthly cost rightsizing estimate meters
disk-overprovisioned-p40 rg-prod-data $204.91 $61.47 P40 LRS Disk, P40 LRS Disk Operations
F-SNAP-OLD · Idle / orphaned resources · Medium

2 snapshot(s) appear at least 30 days old (lower bound from the export)

$43.40
ESTIMATE / month

Recommendation. Compare these snapshots to your backup policy. Delete ones past retention, and move long-term copies to a cheaper vault/tier if they still need to be kept.

Assumption / basis. A snapshot is flagged when its charges appear from the first day of the export through a span of at least 30 days (this export covers 92 day(s)). That span is a *lower bound* on age — the actual creation time CAN'T VERIFY from a cost export and may be older. ESTIMATE equals the current monthly run-rate, i.e. the maximum savings if the snapshot is deleted after confirming it is not required for backup or restore. Monthly figures use the October 2026 run-rate (a complete calendar month in the export).

Confidence. Age is a lower bound from the export window (CAN'T VERIFY exact creation time). The dollar figure is an ESTIMATE of maximum monthly savings if deletion is appropriate.

resource resource group subscription monthly cost observed span days days present first seen last seen
snap-old-data-jan rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $27.90 92 92 2026-08-01 2026-10-31
snap-old-os-2024 rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $15.50 92 92 2026-08-01 2026-10-31
F-PIP-REVIEW · Idle / orphaned resources · Medium

3 public IP address(es) are incurring charges — association CAN'T VERIFY

$11.16
ESTIMATE / month

Recommendation. Review each public IP. Release any that are unassociated and not reserved for a planned failover. Keep IPs that are on a load balancer, VM, NAT gateway, or VPN gateway you still need, and any that must retain a stable address.

Assumption / basis. ESTIMATE equals the current monthly run-rate, which is the *maximum* savings if a public IP is unassociated (or otherwise unneeded) and is released. Association status CAN'T VERIFY from a Cost Management export — Standard public IPs are billed for the hours they exist, attached or not. Confirm in the portal (or Resource Graph) before deleting. Monthly figures use the October 2026 run-rate (a complete calendar month in the export).

Confidence. CAN'T VERIFY association from the CSV. The dollar figure is an ESTIMATE of maximum monthly savings if release is appropriate.

resource resource group subscription monthly cost days present
pip-orphaned-01 rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $3.72 92
pip-orphaned-02 rg-leftovers EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $3.72 92
pip-prod-lb-01 rg-prod-network EXAMPLE DATA - Lumen Trail Production (FICTIONAL) $3.72 92
F-UNTAGGED · Governance · Low

Untagged spend is 5% of the period ($663.28 USD)

No dollar claim

Recommendation. Apply a required tag policy (at least env, owner, and application) and backfill tags on the highest-spend untagged resources listed below.

Assumption / basis. No direct dollar savings are claimed. Untagged resources block chargeback, environment-based shutdown, and ownership of other findings. A resource is untagged when the Tags column is empty, null, or '{}'.

Confidence. Observed from the Tags column in the export.

resource resource group period cost
agw-prod-01 rg-prod-network $310.00
disk-orphaned-std-01 rg-leftovers $124.20
snap-old-data-jan rg-leftovers $82.80
disk-orphaned-std-02 rg-leftovers $78.20
snap-old-os-2024 rg-leftovers $46.00
pip-orphaned-01 rg-leftovers $11.04
pip-orphaned-02 rg-leftovers $11.04

Appendix: method, columns, and sources

How this file was read

Schema detected as: Cost Details / usage details (Exports or Cost Details API, ActualCost or AmortizedCost) — https://learn.microsoft.com/en-us/azure/cost-management-billing/automate/understand-usage-details-fields. Only a date column and a cost column are required; other findings degrade gracefully when optional columns are missing.

Logical fieldHeader used in this fileRequired
additional_info AdditionalInfo no
billing_currency BillingCurrency no
charge_type ChargeType no
consumed_service ConsumedService no
cost CostInBillingCurrency yes
date Date yes
meter_category MeterCategory no
meter_name MeterName no
meter_subcategory MeterSubCategory no
pricing_model PricingModel no
product_name ProductName no
quantity Quantity no
resource_group ResourceGroup no
resource_id ResourceId no
resource_location ResourceLocation no
resource_name ResourceName no
resource_type ResourceType no
subscription_id SubscriptionId no
subscription_name SubscriptionName no
tags Tags no
unit_of_measure UnitOfMeasure no

What each logical field means, and the Microsoft documentation it comes from, is in the README and in the source links below.

Discount ranges cited (Microsoft, retrieved 13 Nov 2026)

Column documentation

What a Cost Management export can and cannot show

Optional later enhancement: read-only API access

A later version could add optional Azure Resource Graph and Cost Management queries (Cost Management Reader + Reader only) to confirm unattached disks, unassociated public IPs, exact snapshot creation times, and Advisor reservation recommendations. That path is not required — a client’s CSV export is the only input this version uses, and no secrets are stored.

Report generated by Oak & Mesa Azure FinOps Audit 1.0.0 from EXAMPLE_DATA_lumen_trail_software_cost_export.csv · 12:54 PM CT, November 13, 2026. EXAMPLE DATA — fictional company, not a real engagement.