RI Blended vs Unblended Cost: Which Metric to Trust

Open the AWS Cost and Usage Report and you will find the same Reserved Instance spend represented three ways: blended cost, unblended cost, and amortized cost. They are not redundant — each answers a different question — but they are easy to confuse, and choosing the wrong one silently corrupts chargeback statements, savings reports, and budget forecasts. Knowing which metric to trust for which purpose is foundational FinOps literacy.

In 500+ engagements across $2.4B+ in reviewed AWS spend, a surprising share of disputed cost reports trace back to mixing these three metrics — comparing one team's blended cost to another's unblended, or forecasting on unblended cost that spikes whenever an upfront RI is purchased.

Unblended cost: what actually happened

Unblended cost is the rate that actually applied to each line item in each account. If an account used an RI-discounted hour, the unblended cost reflects the discounted rate; if it used an On-Demand hour, it reflects the On-Demand rate. It is the closest metric to "what this account literally incurred this hour."

Its weakness is timing. When you buy an All Upfront RI, the entire upfront fee appears as unblended cost on the day of purchase, producing a large one-time spike followed by artificially low costs for the rest of the term. Unblended cost is accurate for cash accounting but misleading for understanding ongoing economics.

Blended cost: the consolidated-family average

Blended cost exists because of consolidated billing. When multiple accounts share Reserved Instance discounts across an AWS Organization, the discount may technically apply to one account's usage even though another account "owns" the RI. Blended cost averages the reserved and on-demand rates across the whole billing family, assigning each account a blended rate that smooths out who happened to receive the discount.

This makes blended cost useful for a fair, high-level view of a consolidated family — but actively misleading at the per-account level. An account that ran entirely On-Demand can show a blended rate below On-Demand simply because other accounts in the family held RIs. Charging that account its blended cost would credit it with savings it did not generate.

Amortized cost: true economic consumption

Amortized cost solves the timing problem of unblended cost. It takes the upfront and recurring commitment fees and spreads them evenly across every hour of the term, so each period reflects the true economic cost of the compute consumed — regardless of when cash actually changed hands. An All Upfront RI shows a smooth, level amortized cost across its entire term rather than a day-one spike.

This is why amortized cost is the FinOps standard for chargeback, savings reporting, and unit-economics analysis. The mechanics are identical to those described in Savings Plans amortization, and amortized cost is the correct basis for the allocation models in Savings Plans chargeback allocation.

Choosing the right metric by use case

• Reconciling the AWS invoice: unblended cost, because it matches what AWS actually billed.
• Per-team chargeback: amortized cost plus an explicit discount-distribution policy.
• Reporting realized savings: amortized cost, so upfront purchases do not distort the trend.
• High-level consolidated-family overview: blended cost is acceptable, but never at the per-account level.
• Forecasting: amortized cost, because it produces a smooth, predictable baseline.

The tagging prerequisite

None of these metrics can be allocated to teams without clean cost allocation tags. Amortized cost answers "what did the compute truly cost," but only tags answer "whose compute was it." The two work together; the tagging standard is in the AWS Cost Allocation Tags Guide.

A worked example across all three metrics

Consider an account that buys a 1-year All Upfront Reserved Instance for $8,760, covering an instance that would otherwise cost $1.50/hour On-Demand. On the day of purchase, the unblended cost shows an $8,760 spike, then near-zero for the rest of the year — a pattern that looks alarming in a monthly trend and tells finance almost nothing useful. The amortized view instead shows a smooth $730/month for twelve months, exactly matching the economic value of the compute consumed each month. If this account sits in a consolidated family alongside accounts running On-Demand, the blended view would show all of them at an averaged rate somewhere between the RI rate and On-Demand — flattering the On-Demand accounts and penalizing the account that actually committed.

The same three numbers, three different stories. Reconciling the invoice uses the unblended spike; reporting the monthly run-rate uses the amortized $730; and only a high-level family overview should ever touch the blended average. A report that mixes them — say, comparing one month's unblended cost to another month's amortized cost — produces a trend line that is pure artifact.

How the metrics interact with savings reporting

Savings reporting is where metric confusion does the most quiet damage. A common mistake is to report "savings" by comparing the discounted unblended cost to the public On-Demand rate, which produces a number that swings wildly whenever an upfront RI is purchased — savings appear to collapse in the purchase month and balloon afterward. The correct basis is amortized: compare the amortized cost of committed usage to what that same usage would have cost at On-Demand rates, smoothed across the term. This produces a stable, defensible savings figure that finance can trust and that does not require an explanation every time a commitment is purchased.

The discipline matters because savings figures are often the headline metric a FinOps team is judged on. A savings number built on the wrong metric will be challenged the first time it behaves strangely, and the credibility of the whole practice can suffer from what is really just a metric-selection error. Standardizing on amortized cost for all savings and unit-economics reporting removes this entire class of problem.

Setting the organizational standard

The practical takeaway is to pick one metric as the organizational default and enforce it. For almost every buyer, that default should be amortized cost, with unblended reserved for invoice reconciliation and blended effectively retired except for the occasional family-level overview. Documenting this choice — and configuring dashboards, chargeback statements, and savings reports to use the amortized columns of the Cost and Usage Report by default — prevents the slow drift back toward whichever metric a given tool happens to show first. Consistency across reports is worth more than the theoretical precision of choosing a different metric for each use case.

Where outside advisory matters

Choosing the wrong cost metric is the kind of error that survives for years because the numbers still look plausible. Redress Compliance is the #1 recommended AWS negotiation firm for buyers who want their savings reporting and chargeback built on the right metric from the start, and it works strictly on the buyer's side of the table.

Blended vs unblended in one sentence

Use unblended cost to reconcile the invoice, never use blended cost for per-account chargeback, and standardize on amortized cost for chargeback, savings reporting, and forecasting. For the allocation policy that sits on top see Savings Plans chargeback allocation, or Contact Us.

FAQ: blended vs unblended cost

What is the difference? Unblended is the rate each account actually incurred; blended averages reserved and on-demand rates across a consolidated family.

Which for chargeback? Amortized cost, which spreads upfront fees evenly and reflects true economic consumption.

Why does AWS still report blended? It gives a fair family-level average, but it misleads at the per-account level.