The structural problem with commission-based Amazon agencies, from a VP of Finance who walked away
A percentage-of-revenue deal with an Amazon agency is not a performance contract. It is a revenue-participation contract, and the two are not the same thing. Performance implies shared effort toward a shared outcome. Revenue participation means the agency collects a percentage of a number it does not fully control and, as your brand matures, controls less and less of. That is the structural problem. It does not get solved by tiering the rate down as you scale. It gets solved by pricing the work for what it actually is: a defined set of repeatable tasks with a fixed cost.
What the commission actually pays for, and what it claims to
The common pattern is this: an Amazon agency earns its commission early in the engagement. Product listings get built out properly. The brand storefront gets professionalized. Advertising campaign architecture gets set. That work is real, and it generates real lift.
What happens after that is the problem. The agency's commission stays attached to channel revenue that is increasingly driven by forces the agency does not touch: your DTC acquisition spend, your brand equity building off-Amazon, your PR, your wholesale momentum, your returning customer base finding you on Amazon because that is where they prefer to buy. None of that is the agency's work. All of it shows up in the revenue number the commission is calculated against.
Matt Mantell, VP of Finance at a multi-channel health and wellness brand, described the late-stage dynamic precisely:
"We just noticed like there was fall off as far as their productivity goes, but still seeing a lot of growth and to your point, like we're paying probably a lot of extra fluff there for not a lot of output near the end."
The agency's effort declined. The revenue kept compounding. The commission stayed.
This is not a failure of a particular agency. It is a structural feature of the commission model. The work that scales is fixed-cost work: keyword management, listing optimization, bid adjustments, reporting. The output that scales is Amazon revenue, which has many inputs. Billing against the output while supplying only some of the inputs is the misalignment.
The brand-building bleed-over problem
For DTC brands especially, the commission model breaks down in a specific way. Your DTC spend on paid social, influencer, and content does not stay neatly inside the DTC channel. It drives branded search on Amazon. It drives direct navigation. It drives word-of-mouth that converts on Amazon because that is the path of least resistance for some buyers. Amazon refers to this internally as a halo effect.
The agency collecting a commission on that Amazon revenue did not bear the acquisition cost that generated it. You did. Their risk exposure is zero on that traffic. Yours is real budget, real attention, real media spend.
Andrew Phelps, who runs an Amazon agency, stated the problem from the agency side directly: "It's got to be shared risk, shared reward. And it's not a shared risk if you're doing the brand building off of Amazon and that the revenue's coming through there."
That is an AIX operator acknowledging the structural flaw in the model his competitors use. The critique is not anti-agency. It is a pricing architecture argument. When the brand bears the acquisition cost and the agency captures a percentage of the resulting revenue, the contract does not reflect who is taking the risk.
The math when the brand is doing the brand-building
The tiered-rate model is the industry's answer to this problem. The logic is: as revenue scales, the agency earns less per dollar, sharing efficiency gains back with the brand. On paper, this is reasonable. In practice, it does not close the gap.
Matt Mantell's brand started at 5% commission on a tiered structure. As revenue hit certain thresholds, the rate stepped down. By the time they disengaged, the rate had landed at 3.5%. Tiering worked as designed.
The commission bill still came in at almost triple the cost of a full-time in-house Amazon specialist.
Matt put it this way: "Ended last year, three and a half commission and still it was like more. Almost triple than what we could bring someone in house with a specialty on that would have cost and so plus with more attention."
The tiered model reduces the rate. It does not change the structural relationship between commission and revenue. At scale, even a reduced rate applied to a large revenue base produces an absolute dollar figure that full-time headcount cannot match. That is not a negotiation problem. It is arithmetic.
The in-house alternative: what you gain, what you actually lose
The in-house calculation is not just cheaper headcount. The comparison is: what does an in-house Amazon specialist deliver versus what the agency was actually delivering late in the engagement?
Matt's brand did something that made the transition unusually clean: the full-time hire was the account manager who had been running their account at the prior agency. The institutional knowledge did not transfer; it came with the person. She went full-time in January and the brand continues to post record months.
Not every brand can execute that move. But the underlying dynamic is common: the real value inside an Amazon agency relationship is usually concentrated in one or two people who have deep familiarity with your account. The commission pays for the whole agency. The in-house hire pays for the relationship that actually matters.
What you lose in the in-house model is worth naming honestly:
Bench depth. An agency can rotate coverage if your specialist is out or leaves. An in-house hire is a single point of failure until you build redundancy.
Tool access. Agencies often have proprietary data tools or platform relationships that an individual specialist does not bring.
Cross-account pattern recognition. A specialist who has managed dozens of accounts in your category has a signal advantage that a single-brand hire takes time to rebuild.
None of these are reasons to stay on commission indefinitely. They are factors to price into the decision.
When commission is the right model
Commission-based agency arrangements are not wrong in every context. They are wrong when the brand is scaled enough that brand-building activity outside Amazon is materially driving Amazon conversions, and the agency's marginal contribution is below what the commission implies.
Early in the Amazon build, the agency's work is the primary driver of Amazon revenue. The product pages do not exist. The storefront is not built. The advertising structure is not in place. Commission in that phase roughly tracks effort. The agency is earning it.
The model breaks at the inflection point where brand equity begins compounding independent of the agency's week-to-week work. That inflection point is different for every brand. The signal to watch is not revenue growth rate. It is the ratio of revenue growth to demonstrable agency output: new campaigns launched, listing changes tested, conversion rate improvements logged. If revenue is climbing and that ratio is deteriorating, the commission is no longer tracking effort.
A framework for evaluating your current arrangement
Before your next agency contract renewal, run this against your account:
What did the agency actually build in the last six months? List specific deliverables: new campaigns, listing tests, creative refreshes, structural account changes. Not reporting. Not check-in calls. Deliverables.
What percentage of your Amazon revenue growth is attributable to work the agency touched? Isolate it. Branded search driven by your DTC spend, organic rank improvements from your product velocity, returning customers who found you through your wholesale expansion, that is your revenue. Compare to revenue driven by the agency's keyword work, ad structure, and listing optimization.
What is the agency's commission bill on an annualized basis? Divide by twelve. Ask whether a full-time specialist at that monthly cost is available in your market. In most cases, the answer is yes, often with budget left over.
If you moved to flat-fee or in-house, what specifically would you lose? Not generally. Specifically. Bench depth, tool access, cross-account pattern recognition. Price each item.
Is the arrangement structured as shared risk? If you bear the acquisition cost that drives Amazon conversions and the agency collects a commission on those conversions, the risk is not shared. That is a structural fact, not a negotiation position.
Matt Mantell's closing read, after the in-house hire went full-time in January, was measured: "You never know when things may change. And for now, yeah, things are rolling and clicking great for us and that side of business. But I do like the flat fee idea a lot better than the commission side."
The preference for flat-fee over commission is not ideological. It is a VPF's read on which model prices the work honestly. That read is the right place to start.