How much AWS Marketplace actually costs your B2B SaaS startup
AWS Marketplace charges a 3% listing fee on public SaaS offers, and as little as 1.5% on large private offers. That's the number everyone quotes. It's also the smallest part of what listing actually costs you.
The real cost shows up in engineer weeks, reconciliation headaches, and a bet you're making about procurement speed. Here's the full math, not just the headline percentage.
What AWS and Azure actually charge
AWS's fee structure is tiered by deal size, not flat. Public SaaS listings run 3%. Private offers scale down as total contract value climbs: 3% under roughly $1M TCV, 2% in the next band, and 1.5% on the largest deals, with renewals also priced at 1.5%.
If you sell through a channel partner using a Channel Partner Private Offer, add 0.5% on top of whatever tier you'd otherwise pay. A sub-$1M private deal routed through a partner costs 3.5%, not 3%.
Azure used to be the expensive option. Microsoft cut its commercial marketplace fee from 20% to 3% a few years back, which means the two platforms are now priced almost identically. The decision between them isn't a fee decision anymore. It's an ecosystem decision: Azure's co-sell program plugs you into Microsoft's own field sales team if you're Azure-native, which AWS doesn't offer in the same way.
The cost nobody puts in the pitch deck
Here's what the 3% number hides: the engineering lift to get listed correctly.
Marketplace integrations need working metering logic, entitlement checks, and tested activation workflows before AWS will approve you. Get the metering wrong and customers get billed incorrectly, which triggers support escalations and, eventually, a churned account that blames your billing, not your product.
Budget two to four weeks of a senior engineer's time for the initial build, not a weekend project. Then budget ongoing maintenance: every new pricing tier or plan change has to go back through marketplace review, and someone on your team now owns tracking entitlement changes and metering accuracy indefinitely.
There's a reconciliation gotcha too. What your sales team reports as closed and what AWS's disbursement report shows can diverge, because some customers negotiate custom payment terms that AWS doesn't fully expose in seller-facing reports. If your sales comp plan pays out on booked revenue, someone needs to reconcile this monthly or you'll be arguing with your own AEs about numbers that don't match.
The worked example that actually matters
Take a $200,000 ACV enterprise deal.
Sold direct, you keep 100% of it, but you're at the mercy of a new-vendor procurement cycle: security review, legal redlines, a vendor onboarding process that can run four to six months and dies more often than founders want to admit.
Sold as an AWS private offer under the $1M TCV tier, you pay 3%, or $6,000, and keep $194,000. In exchange, Forrester's total economic impact research on AWS Marketplace found buyers close 30 to 40% faster when procurement routes through the marketplace instead of a new direct contract, because the buyer's cloud spend commitment already cleared vendor risk review.
Route the same deal through a channel partner as a CPPO and the fee ticks up to 3.5%, or $7,000. But AWS's own data on CPPO deals shows them closing roughly 50% faster and running four to five times larger than equivalent direct-channel deals, because the partner relationship brings budget authority into the room that a cold procurement process doesn't.
So the real comparison isn't 100% versus 97%. It's whether $6,000 to $7,000 is worth buying two to three months of procurement speed on a deal that might otherwise slip a quarter, or die in legal review entirely. On a $200K deal at a seed-stage company, a slipped quarter is often more expensive than the fee ten times over.
When the fee is worth it, and when it isn't
Marketplace makes sense when your buyer already has committed cloud spend they're trying to draw down, when procurement is the actual bottleneck (not your product), and when the deal is large enough that a few points of fee is trivial next to the time saved.
It makes less sense for small-ACV, self-serve-friendly deals where procurement was never the obstacle. Paying 3% and absorbing weeks of engineering time to speed up a deal that would've closed in a week anyway is a bad trade. Marketplace is a procurement-unlock tool, not a universal payment processor.
What to do this week
Before you build anything, pull your last ten enterprise deals and check how many stalled in procurement or legal, not in the sales conversation itself. If three or more did, the fee math above almost certainly pencils out in your favor. If your deals stall in the pitch, not the paperwork, marketplace listing won't fix your actual problem, and you should fix that first.