What a Free Cloud Waste Assessment Finds in 8 Days

Every cloud cost tool promises to find waste, which is why nobody believes any of them until they see it on their own estate. A cloud waste assessment is the answer to that reasonable skepticism: connect read-only, run the platform against your real bills and your real inventory, and see what it finds before anything is bought or changed. The assessment Opslyft offers is free, needs only read access, and returns a complete picture in a median of eight days: what is idle and orphaned and what it costs, which resources are over-provisioned and by how much, how the untagged third of the bill allocates to owners, how much of the estate your commitments actually cover, and what spiked in the last month and why, each finding with an owner's name on it.

This page walks through what the assessment finds, roughly day by day, what access it needs, what you get at the end, and what organizations typically do with the results. The general shape of the waste it finds is in our guide to where cloud waste hides; this page is about finding it on your estate.

Why Proof Has to Come First

Three reasons buyers are right to ask.

• Every tool claims the same 30 percent. The figure is probably true for your estate too, but a claim is not a finding, and a finding is not actionable until it has an owner and a price.
• Native tools have already shown you a list. Trusted Advisor, Azure Advisor, and Recommender produce findings that sit unactioned, which teaches people that findings are cheap. What is expensive is a finding with the context to act on it.
• Allocation is the hard part and nobody demos it. Anyone can show idle volumes. Showing the untagged third of your bill allocated to teams by rules you can audit is the thing to test, and it can only be tested on your data.

The assessment is designed around those three objections: real estate, owned findings, and the allocation done in front of you.

What It Needs

Everything is read-only, scoped to what you choose, and revocable at the end. Opslyft is ISO 27001 and SOC 2 certified; no resource is created, modified, or deleted during the assessment.

How Opslyft Solves It: Eight Days

Day 1: Connect, and see the idle resources within minutes

With billing exports and inventory access connected, Atlas builds the inventory of every resource across every connected platform and flags the idle and orphaned immediately: unattached volumes, idle addresses, snapshots with no source, load balancers with no targets, gateways in empty VPCs, idle databases, empty clusters, and non-production running around the clock, each with its monthly cost. This is the finding Purplle saw within minutes of connecting and saved $200,000 on within the month.

Days 2 to 3: Every finding gets an owner

Ownership comes from tags where they exist and is inferred from account and project structure, naming, and resource relationships where they do not, through Dimension Studio's rules. By day three the idle list is no longer a list; it is a set of owned items by team, sorted by value, which is the form in which it will later become tickets.

Days 3 to 5: The untagged third is allocated

Dimension Studio maps accounts, subscriptions, projects, and metadata to teams and products and splits shared and untagged spend by fixed, proportional, or usage-driven rules, with the remainder reported as its own line. Coverage typically moves from the 60 to 70 percent your tags reach toward 95 percent, and the rules are shown so finance can audit them. This is the part of the assessment most buyers have never seen done on their own data.

Days 4 to 6: Over-provisioned resources, with the metrics

Atlas's recommendations identify over-provisioned instances, databases, and pod requests using memory, storage, and query-time metrics rather than CPU alone, weighted by each workload's purpose and criticality, so the list distinguishes the production primary from the forgotten staging copy.

Days 5 to 7: Commitments measured across the estate

Reservation and savings plan coverage and utilization are measured on the allocated data across every cloud, not per console, which usually reveals coverage well below what the per-cloud reports implied and under-utilized or expiring commitments nobody owned.

Days 6 to 8: Anomalies with causes, and the report

Monitor runs over the last 30 days and surfaces the anomalies with their root causes, the resource, the change, and the owner. The assessment closes with a report quantifying each category of waste, the allocation of untagged spend, a prioritized list of owned findings with recommended actions, estate-wide commitment coverage, and a walkthrough in the platform, including Iris answering your questions about the results in plain English.

A Worked Example

A company with a $720,000 monthly bill across AWS, Azure, and two Kubernetes clusters connects read-only on a Monday. By Monday afternoon Atlas has flagged $38,000 a month of idle and orphaned resources across 34 accounts and subscriptions. By Wednesday every one of them has an owner. By Friday Dimension Studio has allocated the bill from 61 percent tagged coverage to 94 percent, with the shared Kubernetes clusters split by namespace usage and the remainder reported as a $43,000 line with a target. Over the following days Atlas identifies $52,000 a month of over-provisioned capacity with the memory and query-time evidence behind each item, commitment coverage is measured at 39 percent across the estate against the 62 percent the AWS console reported for AWS alone, and Monitor finds eleven anomalies in the prior month, two of which are still running.

The report on day eight shows about $90,000 a month of actionable findings, 12.5 percent of the bill, each with an owner and an action, plus a commitment gap worth roughly $60,000 a month more and an allocation finance can close on. The company turns the findings into 160 tickets the following week; the two running anomalies are stopped that afternoon.

Proof: eight days, and what the first month looks like. Eight days is Opslyft's median time from first connection to a complete assessment with owned findings and allocated untagged spend. Purplle's first month on the platform returned more than $200,000 from idle and underutilized resources surfaced within minutes of connecting; a leading fintech's assessment of a $40 million GCP estate took unallocated spend down by 75 percent. Read the stories: Purplle and the fintech

Use the Assessment If

• You suspect 20 to 30 percent waste and want it proven on your own estate before buying anything.
• Native-tool findings have sat unactioned because they had no owner.
• More than 20 percent of your bill is untagged or shared and you want to see it allocated.
• You run more than one cloud or platform and nobody has one view.
• Commitment coverage has never been measured across the estate.
• You can grant read-only billing and inventory access for eight days.

If your estate is one account that one engineer knows by heart, an afternoon with Trusted Advisor will serve. The assessment earns its place when the estate is bigger than anyone can audit.

The Bottom Line

A cloud waste assessment is the right way to evaluate a cost platform, because it tests the three things that matter, real findings, owned findings, and allocation of the spend tags cannot reach, on your data rather than a demo's. Opslyft's is free, read-only, and returns in a median of eight days a report you keep: what is idle and what it costs, what is over-provisioned and why, how the untagged third allocates, how much your commitments really cover, and what spiked last month and who owns it. Most estates see 5 to 10 percent of the bill in actionable findings on the first pass. What happens next is in our walkthroughs of finding unused AWS resources and closing the loop from finding to verified saving.