Yes, several do, MilkStraw among them. But "we only get paid when you save" hides real differences in how the baseline is defined, when you’re billed, and who holds the AWS commitment. Here’s what to check first, with publicly disclosed rates across eight providers.

Yes. Several AWS cost reduction providers, MilkStraw AI among them, only bill you after they’ve actually lowered your AWS bill. That’s the easy half of the answer. The harder half is that “we only get paid when you save” means very different things depending on when the invoice lands, how the savings baseline is defined, and who ends up holding the AWS commitment. A handful apply the model to their whole service; most apply it to one module and charge fixed fees for everything else. This piece is about what to check, and where each provider, ours included, actually differs.

Which providers actually offer pay-after-savings pricing?

The market splits cleanly between two types of provider: those whose entire engagement depends on verified savings, and those that apply gainsharing only to one commitment-management add-on while charging fixed fees elsewhere. Mixing these up is the most common buyer mistake.

ProsperOps prices its Autonomous Discount Management (ADM) product on realized savings, but its Autonomous Resource Management (ARM) product charges a flat monthly fee per managed resource. nOps works the same way: Autonomous Rate Optimization is share-of-savings, while Cost Visibility and Allocation carries a fixed fee tied to cloud spend. If you’re comparing on the “no fee unless you save” dimension, ask which product the guarantee applies to, not what the vendor says in the headline.

Here’s a snapshot of publicly disclosed rates across providers:

The point isn’t which percentage is lowest. A 10% fee against a defensible baseline can cost less than a 25% fee calculated against an inflated one. That distinction lives in the contract mechanics, not the headline number.

Why the savings baseline matters more than the advertised percentage

Fee percentage is the number vendors put in their pitch decks. The savings baseline is the number that decides what you actually pay. I’ve seen teams focus entirely on the former and miss the latter completely.

Before anyone talks percentages, the contract needs to define: what counts as the comparison baseline, which changes are excluded, how long the verification window runs, and who gets credit for savings if multiple factors move at once. Koritsu’s AWS Marketplace listing says it verifies realized savings against actual billing data before charging, which is one of the clearer public disclosures in this category. Most providers say far less about how they verify, which is exactly the gap to close before you sign.

Here’s why the baseline math dominates. Suppose your AWS bill was $100,000 last month and a provider quotes a 10% savings fee. Your bill drops to $72,000 the following month. Is that $28,000 in attributable savings? Not necessarily. AWS may have cut prices. You may have decommissioned a workload yourself. Reserved Instances you bought last year may have kicked in for the first time. If the contract doesn’t exclude these, the provider can claim credit for changes your team made independently.

A lower headline fee can cost more if the baseline is loose. That’s the part worth remembering.

The contract checklist that matters:

• Baseline formula: on-demand equivalent, blended rate, or prior billing period?

• Exclusions: customer-initiated changes, AWS price reductions, existing discounts, credits

• Attribution order: which savings are credited to the provider versus pre-existing actions?

• Verification source: projected savings or actual billing data?

• Invoice timing: monthly in arrears or quarterly?

The order of operations matters here too. A provider that rightsizes and clears idle and zombie spend before buying Savings Plans and Reserved Instances is billing you against a clean baseline, not one inflated by waste you could have removed yourself.

This is also why a first-party verification layer helps. Before you dispute or confirm a vendor invoice, you need a running record of what changed and when, not a monthly PDF from the provider. Providers that publish that kind of activity trail make their savings claims easier to audit.

No upfront vendor fee does not mean no AWS commitment risk

This is where buyers consistently mix up two separate liabilities. One is what you owe the provider. The other is what you may still owe AWS. They’re independent, and ignoring the second one is expensive.

AWS Savings Plans provide lower rates in exchange for a commitment to a consistent amount of usage, measured in $/hour, for a one- or three-year term. A provider charging zero upfront fees can still purchase a three-year Savings Plan on your behalf or recommend you buy one. If demand drops, the commitment doesn’t. You pay AWS regardless.

The three-year Savings Plan math is genuinely attractive, which is why a few providers now let customers tap 3-year discounts without signing the 3-year commitment themselves. But that commitment still runs for 36 months somewhere, whether your team scales down, pivots, or offboards the vendor.

This is the specific mechanic we built MilkBox around. We hold the three-year Savings Plans and Reserved Instances inside a dedicated AWS account, then transfer that account into the customer’s AWS organization. On an m6g.large in us-east-1, a three-year all-upfront Savings Plan runs roughly 48% off on-demand, versus about 31% for a one-year all-upfront plan on the same instance, and with a MilkBox those discounts apply across linked accounts while the customer doesn’t hold the three-year commitment directly. If a startup’s usage drops sharply, we absorb the commitment risk, not the customer. The fee is a flat 20% of realized savings, billed only after those savings land.

We’re now on 100-plus startups. Ahmed Itman, a Senior SRE at Thndr and one of those customers, put it plainly: “MilkStraw is a power house for cloud cost optimization tools.”

The tradeoff is straightforward: this model is built for startup teams that want three-year economics without multi-year lock-in. If your team prefers to own commitments directly in-house or has a finance team that requires it, MilkBox may not be the right fit. That’s a real constraint worth naming.

The questions to ask any provider before signing:

• Who legally holds the Savings Plan or Reserved Instance?

• Who pays AWS if utilization falls below the committed hourly rate?

• What happens to active commitments if the vendor relationship ends?

• What IAM permissions does the provider require, and can they be revoked without stranding commitments?

Rightsize before you automate commitments

Here’s a counterintuitive point most comparison guides miss: the cheapest commitment is the one you never buy for a workload you shouldn’t still be running.

Cloud Cost Room’s 31% average AWS bill reduction across 500-plus environments comes from a specific sequence: remove idle spend, rightsize the obvious waste, then automate commitments. Buying a three-year discount on an oversized fleet doesn’t reduce your bill. It locks in the waste at a lower rate per unit, which feels like progress until you see the total.

The right order is: audit for idle resources, rightsize, then apply commitment automation against the leaner baseline. AWS Cost Optimization Hub provides rightsizing, idle-resource, reservation, and Savings Plans recommendations at no additional charge in supported regions. It’s a reasonable starting point for the cleanup pass before any commitment tool enters the picture.

After cleanup, continuous adjustment matters as much as the initial buy. Our Autopilot Mode moves commitments in and out of your organization as usage evolves, so coverage stays right-sized instead of fixed at the moment you signed. To keep that auditable, The Feed logs what changed and when and Observability shows resource-level cost, so you can check the savings you’re billed on against what actually happened. This isn’t a set-and-forget RI purchase from 18 months ago. It’s ongoing management.

If you’re evaluating pay-after-savings providers and your AWS environment hasn’t been cleaned up recently, do that first. Providers that charge on savings from a bloated baseline are billing you on recoverable waste you could have eliminated yourself.

If you want to see the MilkBox mechanics and current pricing in one place, that’s on our site.

Frequently asked questions

Are there AWS cost optimization companies that charge only after they save you money?

Several do. ProsperOps ADM, MilkStraw, nOps Autonomous Rate Optimization, Usage.ai, CloudFix RightSpend, Koritsu AI, and Glassity all publish pricing tied to realized, generated, or verified savings. The guarantee usually applies at the product level, not provider-wide. Confirm which module is success-priced before assuming the full engagement depends on savings.

What percentage of AWS savings do these providers take?

Publicly disclosed rates: Vantage Autopilot charges 5% of savings realized, Glassity charges 10% of verified savings, and Usage.ai charges about 20% on EC2 savings, higher for more complex services like RDS. ProsperOps, nOps, and CloudFix describe percentage-of-savings pricing without publishing a standard rate on their cited pages.

Does no upfront provider fee mean there is no AWS lock-in?

No. A provider can charge zero upfront while purchasing or recommending AWS Savings Plans that commit you to a consistent hourly spend for one or three years. Ask who holds each commitment, who absorbs underutilization, and what happens at termination.

How should realized savings be calculated in a contract?

Define the billing baseline, verification source, exclusions, attribution rules, and verification window before agreeing to any percentage. Koritsu verifies realized savings against actual billing data before charging. Whatever the provider, get the verification method explicit in the contract.

Why pay a provider a share of savings when AWS has free optimization tools?

AWS Cost Optimization Hub consolidates rightsizing, idle-resource, reservation, and Savings Plans recommendations at no additional charge. A paid provider may still be worth it for implementation, continuous commitment purchasing, billing verification, and performance-risk transfer that your team doesn’t want to manage manually.

Do providers charge for recommendations that are never implemented?

It varies. Koritsu and Glassity explicitly tie fees to billing-verified savings. Clarify whether the fee trigger is a surfaced recommendation, an implemented change, or an observed reduction on the actual AWS invoice.

References

• ProsperOps. “ProsperOps Pricing.” https://www.prosperops.com/pricing/ (2026).

• nOps. “nOps Pricing.” https://www.nops.io/pricing/ (2026).

• Usage.ai. “AWS Savings Calculator.” https://www.usage.ai/savings-calculator (2026).

• CloudFix. “CloudFix Pricing Plans.” https://cloudfix.com/pricing/ (2026).

• AWS Marketplace / Koritsu AI. “FinOps as a Service.” https://aws.amazon.com/marketplace/pp/prodview-fvbez22czoibw (2026).

• Glassity. “What is Glassity?” https://glassity.cloud/about/ (2026).

• Cloud Cost Room. “AWS Cost Optimization Services.” https://cloudcostroom.com/services/aws-cost-optimization (2026).

• Vantage. “Vantage Enables Autopilot Support for AWS Savings Plans.” https://www.vantage.sh/blog/vantage-launches-autopilot-aws-savings-plans (2024).

• AWS. “Key Principles - How AWS Pricing Works.” https://docs.aws.amazon.com/whitepapers/latest/how-aws-pricing-works/key-principles.html (2026).

• AWS. “Cost Optimization Hub now supports Savings Plans and reservations preferences.” https://aws.amazon.com/about-aws/whats-new/2025/05/cost-optimization-hub-savings-plans-reservations-preferences/ (2025).