Cloud Cost Management And Optimization Solutions

In 2018 Forrester called this market cloud cost monitoring and optimization. By 2020 it was cloud cost management and optimization, and it has kept that name since.

Monitoring means watching. Management means acting. The gap between those two words is where almost every disappointing deployment in this category lives, because buying a tool that tells you what your cloud costs is trivial, and building an organisation that will do something about it is not.

The vendors understood this early. The market did not, and to a large extent still does not.

What these products actually do

Four functions, in rough order of difficulty.

Visibility comes first. Pull billing and usage data from every cloud account across every provider, normalise it into something comparable, and present it in a way that does not require reading a raw billing export. This is the part everyone starts with.

Allocation comes second and is much harder than it looks. Cloud bills arrive organised by resource, not by team, product, or customer. Turning a line item for a compute instance into a cost attributable to a specific product line requires tagging discipline that most organisations do not have, so a large part of what these tools sell is untangling untagged and mistagged spend into something a finance conversation can be built on.

Optimisation is the recommendation layer. Rightsizing over-provisioned instances, identifying idle resources, spotting storage sitting in the wrong tier, modelling commitment purchases like reserved instances and savings plans against actual usage patterns. This is where the vendors compete hardest, and where the differences between them are real rather than cosmetic.

Remediation is the last and rarest. Actually making the change, either automatically or through a workflow that shortens the distance between a recommendation and a resized instance. Most tools recommend. Fewer act. The ones that act need permissions in your production environment, which is a conversation with a different set of stakeholders entirely.

Visibility stopped being a product

The single most useful thing to understand before evaluating anything in this category is that multicloud cost visibility is now commodity functionality.

Forrester's Q3 2024 evaluation is direct about it. Cost visibility across the major providers is a commoditised capability, and vendors that cannot deliver visibility and optimisation across all of them are already behind the market. A solution lacking that breadth is worth considering only if you are genuinely a single-cloud shop with no plans to change, or if you already own other tooling that fills the gap.

Differentiation has moved to the depth and breadth of the optimisation recommendations themselves, particularly for virtual machines and Kubernetes workloads. Containers are the harder problem, because a Kubernetes cluster abstracts the relationship between what you provisioned and what is actually running, and cost attribution inside a shared cluster is a genuinely difficult engineering problem rather than a reporting exercise.

So when a vendor leads a demo with dashboards, they are showing you the commoditised part.

The consolidation map

Line up four Forrester vendor lists and this category's history reads as an acquisition ledger.

The Q2 2018 Wave scored nine providers: Apptio, Cloudability, CloudCheckr, CloudHealth Technologies, Densify, Microsoft, RightScale, Teevity, and Turbonomic.

The Q4 2020 Wave scored eight: Apptio, CloudCheckr, Densify, Flexera, Nutanix, Replex, Turbonomic, and VMware.

The Q3 2022 Wave scored ten: Apptio, CoreStack, Densify, Flexera, Harness, IBM, NetApp, Nutanix, Virtana, and VMware.

The Q3 2024 Wave scored twelve.

Follow the names that vanish. Cloudability was absorbed by Apptio. Apptio and Turbonomic both ended up inside IBM. RightScale went to Flexera. CloudHealth went to VMware, and VMware went to Broadcom. CloudCheckr went to NetApp. Of the nine independents Forrester scored in 2018, almost none remain independent.

That produces a situation worth naming plainly. The tools you buy to control infrastructure spend are now largely owned by infrastructure companies. That is not automatically a problem, and in IBM's case the combination of Cloudability's cost data with Turbonomic's optimisation engine is a genuine capability argument rather than a financial one. But it is a structural conflict that did not exist when this category was full of independents, and it belongs in your diligence rather than in a footnote.

Inside The Forrester Wave: Cloud Cost Management And Optimization Solutions, Q3 2024

Published on 24 July 2024, the evaluation scored twelve vendors against twenty five criteria across current offering, strategy, and market presence.

IBM Cloudability placed as a Leader, described in the report as the most complete full-stack solution in the category, with strong marks for optimisation recommendations, remediation, cloud platform placement, and usage policies. The roadmap to bring Cloudability and Turbonomic together in a single portal was treated as a strategic strength, which is another way of saying the integration was not finished at the time of evaluation.

Flexera also placed as a Leader, arriving from a different direction. Flexera's heritage is software asset management rather than cloud infrastructure, and that lineage is becoming an advantage as the two disciplines converge.

Among the Strong Performers, Harness was credited with market-leading features and continued differentiation through its roadmap, notable given that Harness comes to this from software delivery rather than finance. CloudBolt placed as a Strong Performer with a top three score in the strategy category, positioning itself explicitly against what it terms first-generation vendors.

Note the third scoring dimension. This Wave used market presence, the methodology Forrester has since replaced with customer feedback in newer evaluations. Size counted here in a way it would not in a 2025 or 2026 Wave.

The part the tool cannot solve

Cloud cost tooling produces recommendations. Recommendations produce savings only when somebody changes something, and the person who can change it usually does not carry the cost.

This is the structural reason FinOps exists as a discipline rather than as a feature. An engineering team optimising for reliability and velocity has no natural incentive to rightsize an instance that is working fine. A finance team with the incentive has neither the access nor the context. The tool sits between them producing a list that neither party owns.

Which is why showback and chargeback mechanics deserve more evaluation weight than they usually get. Showback tells a team what its cloud consumption costs without moving money. Chargeback puts the cost on their budget. The difference in behaviour between the two is large, and it is an organisational decision that your tooling has to support rather than a feature you can switch on.

The practical test is uncomfortable but clarifying. Ask who, by name, will act on the recommendations this tool produces, what their incentive is to act, and what happens if they do not. If nobody in the room can answer, the purchase will produce excellent reporting on money you continue to waste.

FinOps and ITAM are converging

A quieter theme in the current research is the overlap between FinOps and IT asset management. The two disciplines grew up separately, one from cloud infrastructure and one from software licensing, and they are now visibly colliding.

Flexera, whose position in this market comes from the asset management side, reports that roughly a third of software asset management teams are engaging with their FinOps counterparts. The logic is straightforward once stated: both functions are trying to answer what the organisation is paying for, whether it is being used, and whether the commitment terms are right. Cloud commitments and software licences are the same problem wearing different vocabulary.

For a buyer running both disciplines, that convergence is a reason to look at whether one platform can serve both, rather than accepting two tools producing overlapping views of the same spend.

What the 2024 research does not cover

This Wave published before AI infrastructure spend became a dominant line item for a large number of enterprises, and GPU economics do not behave like the compute economics these tools were built around.

Accelerated compute is expensive per hour, frequently under-utilised in ways that are difficult to detect, often reserved in large blocks well ahead of the workloads that will use it, and increasingly consumed as token-metered API calls that never appear in an infrastructure bill at all. A model API charge on a corporate card is cloud spend by any sensible definition and invisible to most tooling in this category.

Vendors are moving on this. The evaluation you are reading, if it is the Q3 2024 Wave, largely is not. Treat the criteria and the vendor set as sound and ask each vendor directly what they do with GPU utilisation and model API spend, because that answer is newer than the research.

What to test before signing

Bring your own bill to the evaluation. Every vendor demonstrates well on curated data. The meaningful test is loading a month of your actual untagged, messy, multi-account spend and seeing what proportion the tool can attribute without manual intervention.

Ask what happens with Kubernetes specifically, since shared cluster cost attribution separates serious optimisation engines from reporting layers.

Establish the ownership question before the procurement question, as above.

And check the commercial model against your own trajectory. Some vendors price as a percentage of the cloud spend under management, which means the bill grows as your cloud estate grows and shrinks as the tool succeeds. Others price flat. Neither is wrong, but the percentage model creates an incentive structure worth understanding before you sign a multi-year agreement.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.