The One Question Your Amazon Seller Analytics Never Answer
Table of Contents
Your dashboard shows revenue went up last month. Your ad spend held steady. Units shipped faster than the month before. Everything points in the right direction — until you check the bank deposit and the number does not match.
The gap between what your reports show and what you actually earned comes down to one question: how much did you actually keep? Standard Amazon seller analytics can tell you how much you sold, but they often do not show what remained after referral fees, fulfillment costs, returns, advertising, and COGS. For agencies managing multiple accounts, that blind spot multiplies across every client.
Most Amazon and Walmart sellers track revenue but miss profit. The fix is not more data — it is asking questions that force your numbers to show where money enters, where fees take it, and what stays.
Why this happens
The default reports inside Seller Central and Walmart Seller Center were built for order tracking, not margin analysis. Business Reports shows sessions, page views, and units ordered. The Payments report shows settlements. Neither one places a product’s revenue next to its referral fee, FBA fee, return cost, and COGS in the same row.
So sellers and agencies fill the gap with spreadsheets. They pull a settlement report, cross-reference it with an advertising report, manually add COGS, and arrive at a number they hope is right. The platform splits the answer across reports that were never designed to talk to each other.
Amazon charges a 15% referral fee on most categories: Home & Kitchen, Toys, Sports & Outdoors, Office Products. Electronics and computers pay 8% (Amazon referral fees by category, 2026). FBA fulfillment fees increased by an average of $0.08 per unit in January 2026 (Amazon’s fee update notice). Those two fees alone eat into margin at different rates depending on your category mix.
Storage fees, return processing fees, removal fees, and advertising costs all live in separate reports too. When the question is “how was last month?” and the answer is a single revenue total, the fees that shaped the real outcome stay hidden.
What weak seller analytics actually cost you
Take a product selling at $29.99 in the Home & Kitchen category. The referral fee is 15%, which is $4.50. The FBA fulfillment fee for a standard-size item runs roughly $3.50–$6.00 depending on weight. Use the midpoint at $4.75.
| Line item | Amount |
|---|---|
| Sale price | $29.99 |
| Referral fee (15%) | −$4.50 |
| FBA fee (standard-size, mid-range) | −$4.75 |
| COGS (your input) | −$_. |
| Net before ad spend | $20.74 minus your COGS |
That $20.74 is the ceiling — before advertising, returns, and storage. Plug in your own COGS, subtract your ad spend per unit and your return costs, and the gap between $20.74 and those numbers is what you actually keep.
The question that catches this: “What is my net profit per unit, per product, after every Amazon fee?” A good answer shows the math line by line. A bad answer says “margin is around 25%.”
For agencies, this compounds. A product-level margin question answered with account-level percentages hides which products earn money and which ones quietly drain it. Multiply that across client accounts with different category mixes, and the reporting gap becomes a profit calculation problem no single spreadsheet can keep current.
How to fix it
This is where asking the right questions changes what your analytics actually show you. The data already exists — the problem is bringing it together. AI can help here. Instead of manually searching through multiple reports, an AI system connected to your business data can surface the answer. MCP, or Model Context Protocol, gives AI systems a standard way to connect with external data and tools. But whether you find the answer manually or use AI, the questions themselves still matter.
Ask “where did the money go?” instead of “how much did I make?”
Revenue minus COGS is not profit on Amazon or Walmart. Revenue minus referral fees minus fulfillment fees minus storage fees minus return costs minus ad spend minus COGS is profit. The question dictates what the answer has to include.
Every time you review a product, the answer should show every fee charged again
Ask “which products actually earn money after ads?”
ACOS tells you how much you spent on ads relative to ad-attributed revenue. It does not tell you if the product was profitable after those ads ran. TACOS gets closer, but still leaves out fees and COGS.
The better question: “For each product, what is the net profit after every fee and the full ad spend?” Rank products by profit per unit, not revenue. A top-selling product with thin margins and high ad spend can cost more than it earns. The only way to catch it is a question that forces every cost into the same view.
Ask “what changed since last period — and why?”
A month-over-month revenue comparison is useful. A profit comparison is more useful. But neither explains why the number moved.
Ask: “Which specific cost lines changed, and by how much?” Did the referral fee shift because of a category reclassification? Did FBA fees rise because packaging dimensions changed? Did return rates spike on one ASIN? A good answer isolates the variable. A bad answer says “profit dropped 12%” and stops there.
Ask “am I getting reimbursed for what Amazon owes?”
Amazon loses inventory, damages units, overcharges on incorrect dimensions, and processes refunds where the item never comes back. All of it is reimbursable — within time windows that vary by claim type, some as short as 60 days.
The question: “Which reimbursement-eligible events happened this period, and have claims been filed?” A good answer itemizes each event type with the dollar amount recoverable. A bad answer is silence, because the question never got asked.
Finding a reimbursement case buried in fee and inventory data is not always obvious from the disposition codes alone. KwickMetrics scans each connected account’s ledger for lost, damaged, and overcharged-fee cases and surfaces the ones with enough evidence to file, on Amazon and on Walmart. Either way, the question above is the one worth asking every month.
Doing this across multiple accounts
Everything above gets harder when you manage more than one account. The same questions apply to every client, but answers have to stay separate — different categories, different fees, different COGS, different ad strategies.
For agencies, there is one more question: “Which client accounts are profitable to serve?” That requires each account’s fee-level P&L, not just revenue. AI can help here too — instead of looking at each account separately, you can ask questions that cover your entire portfolio:
“Which of my client accounts are the most profitable?” “Which accounts have declining margins and need attention?” “Where are we losing money across all accounts?”
MCP helps AI connect with the relevant business data and context needed to answer these questions across multiple accounts. Running this manually means pulling the same reports across every account, every period. The fee structures do not simplify when you add accounts — they multiply.
KwickMetrics puts every account’s P&L — down to the fee line, per product, per day — under one login, on both Amazon and Walmart. No per-account fee. The same questions this article walks through are the ones the platform answers by default: net profit per unit, ad-adjusted margin, and reimbursement-eligible amounts, with the math visible. That gives AI the connected data it needs to answer business questions across every client.
Pick your top-selling product. Pull every fee Amazon or Walmart charged on it last month. Add your COGS and ad spend. Subtract it all from revenue. If that number surprises you, the rest of your catalog will too. Run the same check on your next five products, then compare margins across categories. For agencies, repeat it account by account — the products losing money on one client are rarely the same ones losing money on another.
The question is the same every time: what did you actually keep, after everything? Ask it monthly, not once a year, and the catalog-wide picture will already be in view before a quarter goes wrong.
Bring every client account under one login — no per-account fees.
Get Your Questions Answered (FAQ)
What is my net profit per unit, per product, after every fee?" This forces every cost — referral, fulfillment, storage, returns, ad spend, COGS — into the same calculation. Revenue alone does not tell you which products earn money.
Each client account has a different category mix, COGS, and fee structure. Agencies need fee-level P&L per account — not aggregated revenue — to know which clients are profitable to serve. Connected analytics can make this easier by providing a consistent view across accounts.
No. Seller Central and Walmart Seller Center remain the underlying sources of business data. AI can help make that data easier to analyse by bringing relevant information together and helping answer business questions without requiring sellers to manually work through multiple reports.
At minimum, monthly. Products that were profitable last quarter can slip into negative margin when fees change, return rates shift, or ad spend climbs. A weekly check catches problems while you can still adjust.
Referral fees, FBA fulfillment fees, monthly storage fees, long-term storage fees, return processing fees, removal and disposal fees, and advertising costs. Each one is reported separately inside Seller Central, which is why most sellers miss the full picture until they bring every fee into one calculation.
Amazon withholds fees, refunds, and reimbursements before settling funds. The settlement report shows the net transfer, but it does not break down which fees reduced the total or which products drove the gap. Comparing the settlement to your order-level fees is the only way to reconcile the two numbers.
ACOS measures ad spend as a percentage of ad-attributed revenue. It does not account for referral fees, fulfillment costs, returns, or COGS. A product can have a low ACOS and still lose money once all costs are subtracted. Real profitability requires every cost in the same calculation, not just the advertising line.
