A stock keeping unit is the code you own for one sellable version of one product. It is the level every operating number resolves at: cost per unit, return rate, conversion rate, and the four signals that decide Scale / Fix / Kill. Confirm the naming rules inside your own Seller Central account.
The short version
- The code is yours before it is anyone else's. You decide what counts as one sellable version, and that decision is made before anything is typed into a form.
- What you count separately sets what you can decide. Two products sharing one line return one blended number, and a blended number has never killed anything.
- Cost per unit belongs on the line, not on the supplier invoice. Landed cost, fees, and returns have to resolve to the thing sitting in the box.
- A return rate under 8% is a per product reading. Averaged across a catalog it hides the single product quietly eating your margin.
- Four signals decide Scale / Fix / Kill over 60 to 90 days. Rating trend, return rate, conversion rate, and cost of customer acquisition trajectory, read per line.
What a stock keeping unit is, and what it is not
You decide what counts as a version: one size, one color, one pack count, or one bundle. That is a business decision, and it is finished before anyone opens a screen.
What the code is not is a description, a report, or a piece of analysis. A code nobody can read back to a landed cost is a string rather than a record. How the field behaves inside Seller Central is account detail I will not guess at from outside, so confirm it there.
The stages that make one product readable, gate by gate
Run these in order and stop at the first gate that does not clear. A stage without its gate produced no decision, and the next stage inherits the fault.
| Stage | What it proves | The gate to the next stage |
|---|---|---|
| 1. One line for every sellable version | The number of things you sell is a fact, not an impression | You can say the count out loud and your warehouse agrees |
| 2. Landed cost on each line | Cost per unit exists per product instead of per invoice | Every line carries a cost you would defend to a buyer |
| 3. Returns attached to the same line | Return rate reads per product against the 8% ceiling | No line sits above 8% without a named fix and a date |
| 4. Traffic and ad spend attached to the line | Cost of customer acquisition belongs to a product, not a month | Spend maps to lines without an allocation guess |
| 5. The four signals read on the line weekly | Scale, fix, or kill becomes a decision with evidence under it | A written stop rule names the window and the person |
Flapen figures as of September 2026.
Stage three is where most small catalogs break. Returns arrive as a total, the total gets divided across everything, and the one product generating them disappears into an average that looks survivable.
What a muddled catalog costs a 200 unit window
This page is for the seller running one to three products at $5K to $30K a month. That seller usually arrives with one sentence: "My product is live but sales are not where they should be." At that size the catalog is small enough to fix in an afternoon and large enough to hide the answer for a quarter.
Phase 1 commits 200 units and $5,000 to $10,000, with up to 4 products tested at once. Those units are bought to produce three readings and nothing else: rating, conversion rate, and cost of customer acquisition. A catalog that blends four experiments into one line returns one reading, which is the same as returning none.
The window runs 60 to 90 days and it does not stretch because a spreadsheet was untidy. Every week the four signals cannot be read per product is a week of that window spent buying data you cannot use.
So the cost is not the tidying, which takes an afternoon. The cost is the decision you postpone while the units run out.
What a management service will not tell you about your catalog
The same sequence works as a test of anyone you pay to run the account, mine included. Send the five questions in writing and read each answer against the gate column.
| Stage | What it proves about them | The gate |
|---|---|---|
| 1. Ask for your sellable count before the quote | Whether they read the catalog or only the revenue | Their count and your count are the same number |
| 2. Ask what landed cost they would put on each line | Whether they price the work against contribution or sales | They ask for your cost sheet before answering |
| 3. Ask which signals they report per line each week | Whether the weekly report carries a decision or a summary | All four appear per product, never blended |
| 4. Ask what would make them tell you to stop | Whether a stop rule exists before the money is committed | A window in days and a named person, in writing |
| 5. Ask what your fee does when a product is killed | Whether their revenue survives your own discipline | The fee falls with the count, written in the contract |
Flapen figures as of September 2026.
Our fee is indexed to your product count, so one product is $800 a month, two are $1,150, and three are $1,500. Killing a product costs us money, and the tier list is published anyway. Ask every candidate to point at the clause where their invoice moves when your catalog shrinks.
50 operators here read the four signals per product every week, by hand, and the read arrives as a written update in Slack with a live review every two weeks. That cadence is the product.
Hold us to every row above. The contract runs month to month on 30 days' notice, the written audit costs nothing, and on exit you keep the Seller Central account, the campaigns, and the creative. If your own count and cost sheet already clear these gates, run the work yourself and keep the fee.
Related answers
- Amazon feedback
- Amazon ASIN label
- Best agencies for Amazon brand management
- Case studies for scaling revenue on Amazon
- Done-for-you Amazon management: the complete guide
One thing to do this week, at no cost. You run one to three products at $5K to $30K a month, so this fits on one sheet of paper. Write a line for every version a customer can buy, then four columns beside it: landed cost, returns over the last 60 days, conversion rate, and last month's ad spend.
Fill in what you already hold and leave the rest empty. The empty cells are the decisions you have been making without evidence, and the widest column of blanks is next week's work.
Request the free written audit to have that sheet read against your listings and your return rate, with prioritized fixes back inside 48 hours at no charge, from Flapen.
