Shared and platform costs are the spend no single team owns directly, such as commitment discounts, support fees, networking, shared Kubernetes clusters, and central data and observability platforms, and they are allocated by spreading them across consuming teams using a defensible key. The three workable methods are proportional to usage, even split, and a usage based key per resource, and the right choice is the one teams perceive as fair and that ties back to a metric they can influence.
Here is what counts as a shared cost, the allocation methods and when each fits, how to handle the unallocated remainder, and a worked example of a model teams accept.
What counts as a shared cost?
Shared costs are everything the tagging policy cannot assign to a single owner. The big categories are commitment discounts and their unused portion, since a Savings Plan or a set of Reservations is bought centrally and its benefit and risk span many teams; support and enterprise agreement fees such as an EDP or MACC drawdown; networking like NAT gateways, load balancers, and inter zone or egress traffic that serves multiple services; shared Kubernetes clusters hosting many tenants; and central platforms for logging, monitoring, data warehousing, and now AI. Left unallocated, these costs make every team's bill look artificially low and hide the true cost of running a product, which is exactly the visibility chargeback exists to create.
Which allocation method should you use?
There are three defensible methods, and most organisations use a blend keyed to the type of cost.
| Method | How it works | Best for |
|---|---|---|
| Proportional to usage | Split in proportion to each team's measured consumption of the underlying resource | Commitments, shared clusters, networking tied to traffic |
| Even split | Divide equally across teams or products | Fixed platform costs every team benefits from equally |
| Usage based key | Allocate by a specific metric such as vCPU hours, requests, or gigabytes | Granular platform and data costs with a clear driver |
Commitment discounts are best spread proportionally so a team that drove the usage that justified the commitment receives its share of the discount, and its share of any unused portion. Support fees often split evenly or in proportion to total spend. Shared cluster cost allocates cleanly by each namespace's measured CPU and memory usage. The principle is that the key should track something the team can influence, so the chargeback creates an incentive rather than a fixed tax.
How do you handle the unallocated remainder?
No model allocates one hundred percent cleanly. Untagged resources, rounding, and genuinely central overhead leave a remainder, and how you treat it shapes whether teams trust the system. Keep the unallocated pool visible as its own line rather than silently smearing it across teams, set a target to shrink it month over month, and use it as the backstop that drives tagging discipline: when the cost of being untagged is a shared penalty, teams tag their resources. A small, shrinking unallocated line is healthy; a large one that nobody owns is a sign the tagging policy is not being followed.
A worked example
A financial services company ran a shared platform team whose costs, commitments, a shared cluster, networking, and central logging, landed in one bucket that product teams ignored because none of it touched their budgets. Splitting commitments and the shared cluster proportionally by measured usage, splitting support evenly, and keeping a visible unallocated line gave each product a fully loaded cost for the first time. Two products that had looked cheap turned out to carry most of the shared cluster, and once that showed up in their numbers they rightsized their workloads. The allocation itself saved nothing; the visibility it created drove the reductions. Figures are verified against billing data and anonymised.
Frequently asked questions
How do you allocate shared cloud costs fairly?
How should commitment discounts be allocated?
What do you do with costs you cannot allocate?
Build an allocation model teams accept
We design shared cost allocation that is fair, transparent, and consistent, so chargeback drives reductions instead of arguments, as an independent advisory that takes zero provider commissions and answers only to you. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on a Fixed Fee or a no risk Gainshare basis. Download the cloud cost optimization playbook, read the deeper FinOps operating model guide, and on the backstop see untagged spend, the allocation backstop. For monthly buyer side analysis, subscribe to The Cloud Spend Navigator.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.