$21M Pet Brand on Amazon — What Did They Actually Keep?

An agency shared a transformation story publicly. A pet brand doing $60,000 per week with TACOS around 30%. Two years later: $400,000+ per week. TACOS: 5.2%. Monthly revenue past $1.7 million.

The dashboard confirms it. $415,000 in the last seven days. Roughly $21.6 million annualised.

The seller's story is about ad efficiency — fixing campaign structure, reducing cannibalisation, reallocating budgets. Revenue grew 6.9× while ad spend only grew 1.2×. The optimisation narrative is compelling.

But here's the question nobody asked: what happened to the platform fees?

The fee stack that didn't change

Every number below is from published Amazon rate cards. We don't have this seller's ASP, so we're assuming $30 per unit — a working midpoint for pet products. That gives us roughly 59,900 units per month.

Fee	Per unit	Monthly	Rate card
Referral fee (15%)	$4.50	$269,543	Source
FBA fulfilment (Large Std, 8–12 oz)	$4.20	$251,573	Source
Inbound placement	$0.30	$17,969	Source
Professional plan	—	$40	Source
Platform total	$9.00	$539,125

Platform take: 30.0% of revenue. It was 30% when the brand did $60K per week. It's 30% now at $400K per week. The seller optimised everything they could control. The platform's cut didn't move.

The cascade: what happens when TACOS drops

This is where the transformation gets interesting. The seller cut TACOS from 30% to 5.2%. At current revenue, that's a difference of $445,644 per month in advertising costs.

	TACOS 30% (old rate)	TACOS 5.2% (current)
Monthly revenue	$1,796,950	$1,796,950
Platform fees (30%)	−$539,125	−$539,125
Advertising	−$539,085	−$93,441
After fees + ads	$718,740 (40.0%)	$1,164,384 (64.8%)

Same revenue. Same platform fees. The only variable that changed was ad efficiency — and it moved $445,644 per month to the bottom line.

But look at what that means in reverse. Even at 5.2% TACOS — an exceptional number — the platform still takes 30%. The seller optimised their largest controllable cost (advertising) from 30% down to 5.2%, and the platform's share remained six times bigger than the ad spend.

The ad optimisation is real. The platform fees are the fixed constraint underneath it. No amount of campaign restructuring changes the referral rate or the per-unit fulfilment fee. Those costs are structural. The only question is whether there are published alternatives within the fee structure that reduce them — and whether the seller knows about those alternatives.

What if the fee stack moved too?

The seller demonstrated that changing one input — ad efficiency — moves the entire waterfall. So what happens if a fee input moves?

Weight reduction. If the average product shifted from the 8–12 oz band ($4.20/unit) to the 4–8 oz band ($3.95/unit), fulfilment costs drop by $0.25 per unit. Across 59,900 units per month: $14,975/month — $179,700/year. (Source)

SIPP certification. If products can ship in their own packaging, the SIPP discount for Large Standard 8–12 oz is $0.07 per unit: $4,193/month — $50,316/year. (Source)

These numbers look small next to the $445K monthly ad savings. They're not. The ad optimisation was a two-year project involving campaign restructuring, keyword harvesting, inventory planning, and listing optimisation. The fee reductions are structural changes that take effect on every unit, in perpetuity, with no ongoing management.

$230K per year from weight reduction and SIPP — applied once, no campaign management required.

The seller spent two years fixing their advertising. The fee structure was sitting there the whole time.

The classification line nobody checks

Pet Products on Amazon has two referral fee rates. Most sellers only know about one.

Classification	Referral rate	Monthly referral (at $1.8M revenue)
Non-veterinary-diet products	15%	$269,543
Veterinary diets	22%	$395,329

(Referral fees — verified 2026-04-18)

The difference: $125,786 per month. $1.5 million per year.

If any SKU in this pet brand is classified as a veterinary diet — and at this scale, with a full catalogue of pet products, the classification boundary matters — the referral fee on that SKU jumps from 15% to 22%. Seven percentage points. On a single classification field.

This is the kind of fee line that doesn't show up in any PPC audit. It doesn't show up in a TACOS analysis. It doesn't show up in a campaign restructure. It sits in the product classification, applies to every unit from the first sale, and most sellers never check it because they don't know the second rate exists.

The seller spent two years optimising $445K per month in ad savings. A single misclassification on one product line could cost $125K per month — wiping out three months of that work.

The dashboard gap

The Seller Central app shows $415K in the last seven days. It shows daily sales, ad impressions, coupon redemptions. It doesn't show:

• Which units are paying 15% referral and which might be paying 22%
• What the per-unit fulfilment cost is across different product weights
• Whether SIPP certification would reduce costs on any SKU
• How the fee stack changes if pricing shifts above or below the $50 band
• What the coupon sales ($58,702 in the last 30 days) actually cost after the 2.5% variable fee and the funded discount
• Whether the B2B orders the seller processes qualify for multi-unit fulfilment discounts

The dashboard shows revenue. The fee stack underneath it requires a different kind of computation — one that models every fee, on every product, through every classification, and shows what moves when you change an input.

The gap

Every seller in this series optimised something — ads, pricing, product mix, margins. None of them computed the full fee stack underneath.

This seller proved that moving one input (ad efficiency) can shift $445,000 per month. The fee structure has dozens of inputs. Most of them have published alternatives. Almost none of them get checked — because the tools sellers use show revenue and ad metrics, not fee structures.

$230,000 per year from weight reduction and SIPP certification — applied once, no ongoing management. $1.5 million per year at risk from a single category classification field. Both sitting in the published fee structure. Neither visible in the advertising dashboard the seller spent two years optimising.

The platform's 30% doesn't change by accident. It changes by computation.

---

Every fee computed from published Amazon rate cards. Sources linked inline. Rates verified 2026-04-18 to 2026-06-15. Last updated: 2026-06-24.

Want to model what happens when you change an input — weight, classification, fulfilment method — across every product? Get early access →