Operations, fees & money
How Amazon Works · C6
FBA vs FBM, inventory, fees, tariffs and refunds, unit economics and CAC — the arithmetic underneath every other chapter.
FBA vs FBM, inventory, fees, tariffs and refunds, unit economics and CAC — the arithmetic underneath every other chapter.
Amazon is a fee machine before it is a marketplace: its average take of a third-party sale crossed 50 % in 2022 [L10], and the sellers who survive know their margin to the dollar before spending on ads or inventory. Practitioners converge on one shape — a third of the price to Amazon, a third to landed cost, a third as profit [L83] — and one rule: your break-even advertising ratio is your margin after every other cost [L32]. Everything else here (fulfilment choice, price bands, restock timing, refunds you must claim yourself) is about protecting that third.
1. Where the money goes: the fee stack
Every unit pays a fixed set of lines; sellers who model only referral and FBA fees are the ones surprised later.
| Fee line | What drives it | What the sources say |
|---|---|---|
| Referral fee | % of price, by category | ~15 % in most categories, same for FBA and FBM [L83]; vs TikTok Shop ~11 %, Etsy ~9.5 %, ChatGPT checkout ~7 % [L13] |
| FBA fulfilment fee | size tier + weight (dimensional if larger) | flat per unit under FBA; FBM prices its own box + carrier [L83]; dimensional weight above actual is a profit red flag [L94] |
| Storage | cubic feet × month, seasonal | model with the Q4 worst case [L83]; slow movers rack up long-term fees [L86] |
| Aged-inventory fee | units past an age threshold | liquidate the aged units, not the whole ASIN [L80] |
| Low-inventory-level fee | days of supply vs velocity | waived for retiring SKUs (<~20 units/7 days), 4-week grace after Prime Day, refunded when Amazon's inbound delays caused it [L80] |
| Inbound placement fee | how few warehouses a shipment splits across | cut by splitting to more sites or ≥5 identical cartons per item [L50] |
| Surcharges & payout | policy changes | 3.5 % fuel surcharge from 17 April 2026; payout clock now 7 days after delivery, not shipment; ad billing from earnings from 1 August 2026 [L10] (US — verify for .ae) |
Two mechanics matter more than the rate card. Amazon measures the unit itself — the cubiscan reading on first receipt overrides your declared dimensions, and shaving half an inch to an inch off packaging can drop a size tier once you sell through old stock and request a re-scan [L80]. And oversized packaging can push per-unit shipping above the product's own landed cost [L73].
2. FBA vs FBM — the ten reasons and the four exceptions
FBA (Fulfilled by Amazon: you ship stock into Amazon's warehouses; Amazon picks, packs, ships and handles customer service) is the default; FBM (Fulfilled by Merchant: you ship each order) is the exception. One audit called a top listing not on FBA "a dealbreaker" to fix before anything else [L72].
| Why FBA wins [L86] | Figure |
|---|---|
| Prime badge lifts conversion | +5–10 % vs non-Prime |
| Featured-offer (buy box) eligibility | rules changed to structurally favour FBA regardless of seller type |
| Amazon absorbs customer-service risk | order-defect rate, negative feedback, A-to-Z and chargeback claims stop hitting account health |
| Mobile Prime filter | applied by default for the >70 % of buyers on mobile |
| Cheaper shipping at scale | ~$3+/unit FBA vs $4.50–8 via UPS/USPS; a $2 gap × 1,000 units = ~$2,000/month |
| One inventory pool for every channel | Multi-Channel Fulfilment (MCF) ships Shopify/Walmart/TikTok orders from FBA stock |
MCF is the recurring "free money" finding in the Helium 10 audits — brands hand-shipping website orders while holding FBA stock [L72], or keeping a high free-shipping minimum MCF would remove [L69]; one seller tests Walmart.com through MCF instead of a second network [L64]. It costs more than a 3PL but is more reliable and scales for spikes [L80].
When not to use FBA [L86]: oversized/bulky products (inbound and storage erode margin even via AWD, Amazon's bulk-warehousing tier), slow movers (long-term storage fees), high-value fragile goods (Amazon "breaks and loses" them; returns run higher), and sellers already running a large 3PL. Even then the recommended endpoint is a dual-front system: most stock in FBA for the badge and conversion, an FBM reserve so the listing never stocks out when FBA depletes [L86]. Seller Fulfilled Prime is "very, very tricky to maintain" — one agency would "not touch" it [L86], though a beverage seller did run Merchant Fulfilled Prime on a two-day nationwide promise [L64]. Above FBA, AWD suits high-velocity SKUs, a 3PL suits slow or seasonal stock [L09].
3. Unit economics: the arithmetic under every other chapter
Margin targets. Amazon's Revenue Calculator — fed your packaged dimensions and weight rather than a look-alike ASIN — returns referral, fulfilment and storage lines; you add landed cost, and the healthy result is ~33 % net margin, a 100 % return on landed cost [L83]. Reverse it for a COGS ceiling: price − target profit − Amazon fees = the most you may pay a supplier; anything above means another supplier or product [L83]. The Alibaba list price is not that quote [L83].
Where sources disagree: the floor depends on who is speaking. A UK launch playbook demands £7+/unit, 80 % ROI and 30 %+ margin before PPC [L59]; a Data Dive scorecard wants $8–12+ margin [L94]; a first-product seller accepted ≥$5/unit [L49]; buyers of Amazon brands call 15 %+ contribution margin after ads "good", 20–25 % "very healthy", and cut SKUs under 10 % [L90]. Use the pre-ad thresholds [L59] [L83] for go/no-go and the post-ad 20–30 % band [L90] as the steady-state target — they measure different stages.
Price floors. Under $15–20 the fees leave nothing [L42]; under £20 "it can be quite difficult to make a profit" [L59]; a ceiling under $20, weight over 1.5 lb or a 20–30 %+ return rate are each a red flag alone [L94]. High-ticket products need fewer units for the same profit [L42]; one bar for deserving a team's attention is $150–250 profit per day [L79].
Break-even advertising. ACoS (ad spend ÷ ad-attributed sales) above your margin loses money on every advertised sale [L47]; TACoS (ad spend ÷ total sales) is the north-star, and its break-even value is your margin after COGS, shipping, returns, FBA and referral fees [L32]. Max daily ad budget = break-even TACoS × trailing-30-day average daily revenue [L32]; with no history, (CPC ÷ conversion rate) × target daily sales [L32], or an assumed ~50 % TACoS in month one [L19]. A healthy launch shows TACoS falling — 21 % to 6.5 % by week 12 at ~$12k/week [L30] — the move from Launch to Harvest [L19]. Detail in [C3].
CAC. Amazon hides acquisition cost, but blended CAC = ad spend ÷ (total orders − repeat units), from Business Reports (Units Ordered, not Total Order Items) and Brand Analytics' Repeat Purchase Behavior; the worked example gave ~$4.46 [L88]. Brand Analytics counts "repeat" within 90 days; Advertising calls anyone without a purchase in 365 days new-to-brand [L88].
LTV is the biggest margin lever. Taking a customer from one to two purchases a year beats supplier, packaging and shipping negotiation combined [L41]. Repeat share tells you what you are: 20–50 % of sales from repeat buyers is a consumable, ~8 % is not [L41]; consumables should judge ads on lifetime ACoS, which can justify a 90–100 % headline ACoS on new-to-brand campaigns [L41]. Subscribe & Save subscribers spent ~$300 lifetime vs $49 for non-subscribers, yet the S&S price barely differed from one-time — the fix was a real 5–10 % discount plus a first-order S&S coupon, aiming for 30–35 % of daily orders via S&S [L68] [L71]; another operator uses S&S coupons only to lift conversion while staying month-to-month profitable [L61].
SHIO unit-economics template (filtered shower head)
Fill with amazon.ae figures, not US rates ([C8] says where to look).
| Line | Symbol | Where it comes from |
|---|---|---|
| Selling price | P | your live listing; check it is not in a dead-zone band (§4) |
| Landed cost per unit | C | factory + freight + UAE duty/VAT treatment + inbound (DDP quote if you can) [L59] |
| Referral fee | r × P | .ae rate card for the category (US analogue ~15 % [L83]) |
| FBA fulfilment fee | F | .ae fee schedule for the size tier after cubiscan [L80] |
| Storage + aged/low-inventory allowance | S | Q4 rate × expected months on shelf [L83] |
| Returns allowance | Rt | return rate × (refund cost + non-resalable share) [L32] |
| Margin before ads | M = P − C − rP − F − S − Rt | this is your break-even TACoS as a % of P [L32] |
| Break-even ACoS / TACoS | M ÷ P | any keyword above it loses money [L47] |
| Target steady-state TACoS | ~6–15 % | one launch settled at 6.5 % [L30]; set your own from M |
| Ad-affordable daily budget | (M ÷ P) × avg daily revenue | ceiling, not target [L32] |
| Sanity check | M ÷ P ≥ 30 % pre-ads; ≥ 20 % after ads | [L59] [L90] |
Run the same table for the replacement cartridge [SS-06]: it is the consumable, so its repeat share and lifetime ACoS [L41], S&S uptake [L68] and its role as "gateway" first purchase in a DSI-style pass over your own shipment data [L09] decide whether the head can run thinner.
4. Pricing mechanics
Dead zones. Because FBA tiers and referral cliffs step up at set prices, there are bands where a higher price yields lower gross profit: for 2026 US categories, baby/beauty/health/personal care ~$10–$11.78, grocery ~$10–$10.88 and $15–$16.23, the general 15 % category ~$10–$10.95/$11.18 plus $50.01–$50.31 [L13] (US — verify for .ae; the mechanism transfers, the numbers do not).
Launch pricing runs in two stages [L47]: undercut the top-20–30 first-page average by ~20 % ($16 vs $20) or price at break-even (COGS + shipping-to-Amazon + fees) to offset zero reviews; then climb ~10 % or $1–2 per week or per 10 new reviews without losing the buy box. UK variant: ~£1 every 5–7 days while rank holds, tapering PPC as rank stabilises [L59]. Launching at a loss with steep coupons is called normal at zero reviews [L73].
Where sources disagree: one seller launches at full price and adds a discount only a month later, banking full-margin organic sales first [L64], against the undercut-first playbooks [L47] [L59] [L73]. Undercutting suits a commodity niche where reviews are the only differentiator; full-price-first suits a product with a real point of difference. Pick by how much you need reviews to justify price.
Badges are engineered from price history. The strikethrough "typical price" shows only when Your Price sits below the median actually paid over the trailing 90 days (promos excluded), so keep list price high and discount via time-limited deals [L47]; a variant raises price ~10 % at a time to reset the "was" baseline [L19]. Coupons need a tick at checkout — many forget and pay full price — and a per-customer cap [L47]; deals and coupons earn mobile real estate even at a ~60 % claim rate [L19]. Splitting one discount into 2–4 stacked pieces beat a single discount of equal value across six experiments and 9,000+ deal posts, strongest at 15–35 % off [L08]. Size Brand Tailored Promotions so the discount ≈ target ACoS [L19]; model any deal (discount, price, COGS, fees, TACoS) before committing [L19]; prepare event pricing in advance [L47] (US event calendar — verify for .ae).
5. Inventory, restock and cash
Sizing the order. Hold ≥100 days of cover because a China-to-US cycle (production + freight + customs + FBA check-in) runs ~100 days; budget = days × daily units × unit cost, e.g. 100 × 25 × $5 ≈ $4,000 / ~800 units [L73]. A $10k launch split ~40 % inventory, ~15 % shipping (~$2/unit), ~15 % branding/photography, ~20 % launch ads, ~5–10 % buffer [L73]; tariffs added ~30 % to the inventory line [L73] (US — verify for .ae). MOQs run 500–1,000 units [L73]; a UK playbook orders ~3 months and reorders 1–3 weeks into launch [L59]; matching 500+ units/month implies a 1,500–2,000-unit first batch [L55].
Restock signals. The FBA Inventory Report (Fulfilment > Inventory > "more") holds days-of-supply and inventory age; watch at ~42 days, act below 28, never fall under ~35 [L80]. Sheets-based planners turn lead time, case size, restock frequency and seasonality into reorder quantities from live inventory and inbound status [L99].
Shipping plans have traps [L50]: unused plans silently eat FBA capacity (the hidden cause of false "exceeds capacity" errors); after confirmation only tracking, carrier, window and quantities up to 5 % or six units per SKU stay editable; plans auto-close after 90 days; the partnered carrier refunds only within 24 hours; mixing up unit and carton dimensions inflates fees. Label units with the FNSKU over any UPC [L50] [L59].
Sourcing terms and payments. Ask for DDP (duty, freight, VAT bundled) so quotes compare and no EORI or forwarder is needed; EXW/FOB mean you arrange freight [L59]. Inspect ~100 units before the balance payment; write every term into the Alibaba Trade Assurance order before the deposit [L59]. Paying via Wise (£4,393) beat card (£4,552) and domestic transfer (£4,415) on one order with identical Trade Assurance cover [L57]. Forwarders quote a "black box": a reverse-auction platform saw $0.66/kg DDP China-to-US vs a ~$1.30/kg market average [L95]. One veteran credits domestic manufacturing for tariff immunity [L78] (US — verify for .ae).
Cash. The payout clock now starts at delivery, not shipment [L10] (US — verify for .ae) — cash flow "shortening from 90 days to zero," one seller said [L10]; buyers price 12+ months of clean P&L and cut sub-10 % SKUs to free capital [L90]. The operator of a $500M brand notes cost levers — shipping rates, customer service, return rate — have lately moved EBITDA more than growth levers [L62]; and external traffic belongs on your own site, where margin is higher [L64].
6. Money that leaks: reimbursements, refunds, tariffs
Amazon does not reliably self-reimburse. Sellerboard's reimbursement-gap report compares Amazon's payout for lost/damaged units against your real cost of goods; Amazon's data reaches back only 60 days, so it must be a recurring check [L98]. File under Fulfilment by Amazon > Other issue with the COGS invoice and a sales report showing the sold price, and cover the other three leaks: lost & damaged, refunded-but-never-returned, overcharged FBA fees [L98]. Low-inventory fees caused by Amazon's own delays are refundable on request [L80].
Tariff refunds are a US story: CBP's CAPE system opened 20 April 2026 to repay ~$127bn of IEEPA duties struck down by the Supreme Court; only the importer of record or its broker files, and refunds take 60–90 days [L10] (US — verify for .ae). Section 321 (<$800 parcels duty-free) is likewise US-only [L80]. For the UAE the question is how duty/VAT enters landed cost — [C8].
7. Account plumbing: identifiers, brand registry, trademarks
- Account tier. Individual caps at 35 units/month and cannot run ads; open Professional once the company and business bank account exist [L59].
- Barcodes. Buy a GS1 GTIN to create the listing, then label units with Amazon's FNSKU [L59] [L50] — or apply for GTIN/UPC exemption ("I don't have a product ID"); approval is per category, brand-approved sellers got it in almost any category, and the form wants a standalone logo plus ≥2 unretouched photos of permanently branded product/packaging [L58].
- Brand Registry accepts a pending trademark number plus branded photos and unlocks A+ Content and Vine [L59] [SS-07]. Register the mark in every marketplace you sell in and defensively in China [L90].
- Gating and compliance. Gated categories need documents (e.g. a Certificate of Analysis) [L73]; a test listing before ordering inventory surfaces gating, patent or compliance blocks [L42].
- Accounting. An Amazon-specialist accountant on a connector such as Link My Books [L59]; Seller Central now integrates with QuickBooks [L13].
- Account health. Under FBA, order-defect rate, negative feedback, A-to-Z and chargeback claims are Amazon's problem [L86]; Helium 10 Alerts flag unauthorised listing changes and new negative reviews [L59].
What this means for SHIO
- Do the sheet before the next ad dirham. Fill the §3 template with .ae fees ([C8] says where) and your DDP landed cost. If margin before ads is under ~30 % [L59], fix price, packaging or COGS before touching PPC — otherwise every campaign in [C3] spends money you do not have.
- Stay FBA, hold an FBM reserve. A shower head is standard-size, mid-value, not glass-fragile — none of the four FBM exceptions apply [L86]; SHIO LLC sits in Dubai, so a small FBM reserve is cheap stockout insurance [L86]. MCF for salon/spa orders is described for US channels — verify on .ae.
- Measure the box. Get the cubiscan dimensions and test whether a slimmer carton drops a size tier [L80]; empty air is a fee [L73].
- Check the price band. Find the .ae fee-tier and referral cliffs and sit just under one, not just over [L13].
- Treat the cartridge as the business. Model the refill's repeat share and lifetime ACoS [L41]; set a real 5–10 % S&S discount plus a first-order S&S coupon [L68]; virtual-bundle head + refill under the buy box [L71]; let refill LTV justify a thinner margin on the head.
- Launch pricing. With a real point of difference (pressure boost + vitamin-C filter), full-price-first with a discount a month later [L64] is defensible; if reviews are thin, undercut ~20 % and climb every 5–7 days [L47] [L59]. Build the strikethrough from a high list price via time-limited deals; cap coupons per customer [L47].
- Restock discipline. ≥100 days of cover from actual daily velocity [L73]; act at 28 days of supply [L80]; delete stale shipping plans [L50].
- Monthly leak audit. Reimbursement gap, lost/damaged, unreturned refunds, fee overcharges — 60-day window [L98]; low-inventory fees Amazon caused [L80].
- Plumbing. Professional tier, GS1 + FNSKU on units, Brand Registry (pending TM suffices) [L59], UAE-plus-China trademark [L90].
- Don't import US tariff/refund tactics (CAPE, Section 321) [L10] [L80] — they do not exist for a UAE importer.
The concept graph behind this chapter
While building the corpus, lore filed the ideas above as pages of a knowledge graph — each link opens that concept's own page, with its definition, the videos it came from, and how to apply it. The 36 most relevant of the subject's 385 published concept pages:
Day Parting (Amazon PPC) · Amazon Profit Red Flags · AWD (Amazon Warehousing and Distribution) · Amazon Seller 10-Step Launch Process · Amazon FBA Exit Preparation Checklist · Amazon as the True Bottom of Funnel · "Amazon Escape" 23-Criteria Product Funnel · Amazon FBA Business Sale Process Stages · Amazon Brand Storefront as Ad-Free Conversion Destination · Five Reasons Amazon FBA Products Fail · Advisor-Led vs. DIY Amazon Brand Sale (Fee Alignment & Risk) · Amazon Customer Acquisition Cost (CAC) Calculation Method · Amazon Gated-Category Documentation Requirement · Amazon Virtual Bundles (Buy Box Cross-Merchandising) · Amazon Product Life Cycle (Launch, Expansion, Harvest) · Amazon Profitability & Real-Cost Validation Toolkit · Rule of Thirds (33% Net Margin) Amazon Profitability Heuristic · Customer Lifetime Value (LTV) as the Primary Amazon Margin Lever · Financial vs. Strategic Buyer Taxonomy (Amazon Brand Sales) · Minimum Deal Size for Buyer Interest (Amazon Brand Sales) · Amazon Creator Connections (Influencer Affiliate Program) · Amazon-Only vs. Omnichannel Seller Skillset Framing ("Inside vs. Outside the Box") · Specialist-Partner Team Structure (Sourcing + Meta + Amazon) · Per-Unit Shipping Cost Gauge (~$2/Unit Heuristic) · Inventory Budget Formula (Days × Daily Sales × Unit Cost) · Preemptive SKU Quantity Configuration for Variable-Quantity Products (Amazon Parent-Child ASINs) · Brand Domination Check · 40-Step FBA Launch Framework · Strikethrough Price & Coupon Badge Engineering · Brand-Family Line Extension for LTV · Two-Stage Launch Pricing Strategy · Order-by-Spend Campaign Audit · Minimum Daily Profit Benchmark (Product Go/No-Go Threshold) · Jungle Scout Extension New-Seller Price Benchmarking (Sales & Review Filters) · Package Size vs. Shipping Cost Tradeoff · Personal Order-History Mining for Product Ideation
Source legend
Not covered / open questions
- No .ae fee schedule here. Referral %, FBA tiers, storage, low-inventory and placement fees for amazon.ae, and whether the fuel surcharge / delivery-based payout clock apply — pull from Seller Central .ae ([C8]).
- Listing hijacking and counterfeits appear only as a fear that stops sellers launching [L08]; nothing on detecting or removing a hijacker, or on Transparency / Project Zero.
- Account-health mechanics (ODR thresholds, IPI, capacity limits, appeals) are named only in passing [L86]; no masterclass exists in these 100 videos.
- Return-rate management for a plumbing-adjacent product (installation, thread compatibility): thresholds exist (20–30 % is a red flag [L94]), no playbook.
- MCF, Subscribe & Save, Brand Tailored Promotions, coupon and deal types on .ae — availability and behaviour to verify.
- Duty/VAT for a Dubai LLC importing into the UAE, the right incoterm, and month-by-month working capital for a single-SKU brand — the corpus is UK/US-shaped and never models it.