An item earns a slot when its market clears $2M per year and you can build 0.2 stars above the niche average. Then 200 units on $5,000 to $10,000 prove rating, conversion rate, and cost of customer acquisition. Four signals decide when the item leaves the catalog, read across 60 to 90 days.
The short version
- The market decides the item, not the item itself. A market under $2M per year cannot repay the cost of acquiring its customers, so nothing under the floor gets a slot.
- The product bar is 0.2 stars above the niche average. Negative reviews on the items already selling name the complaints, so the market writes your brief.
- Every new item enters on 200 units. Phase 1 spends $5,000 to $10,000 to prove rating, conversion rate, and cost of customer acquisition.
- Returns above 8% end the argument. A return rate that high erodes margin whatever the listing does, and the repair sits in the product.
- Four signals decide when an item leaves. Rating trend, return rate, conversion rate, and cost of customer acquisition trajectory, read across a window set in advance.
Score an item before it earns a slot
Most sellers ask which items sell well in FBA. That is the wrong question, because an item inherits its economics from the market it sits inside. Ask whether that market is growing and whether you can capture its traffic profitably.
Score the market on the bars below before a supplier hears from you.
| Criterion | The bar | Weight |
|---|---|---|
| Market size | $2M per year, minimum | Must-have |
| Growth trajectory | Growing year over year | Must-have |
| Return rate | Under 8% | Must-have |
| Rating gap | Room to build 0.2 stars above the niche average | Must-have |
| Traffic you can capture | One channel that pays for itself | Must-have |
| Conversion rate potential | Enough to hit your target cost of customer acquisition | Strong preference |
Flapen figures as of September 2026. Score each row pass or fail in a column of your own.
Write your pass mark down before you score anything. Mine reads as one sentence: every must-have clears, or the item never gets a purchase order. A mark set after the scoring is a preference.
That bar rejects most of what looks good on a research screen. We have scored 193,753 niches at the 2026-08-26 capture, and 4.8% of them passed. So rejection is the normal outcome, and a short catalog is evidence that somebody scored.
Scale, fix, or kill, the review that empties a catalog
A seller running one to three products at $5K to $30K a month says the same thing to me. "I'm spending money on ads but don't know if it's working." Under that sentence there is usually one item that should have left the catalog two quarters ago.
Every live item faces one of three calls: scale it, fix it, or kill it. Four signals make that call.
| Signal | What it proves | The call it drives |
|---|---|---|
| Rating trend | Whether the item is getting better in customers' hands | Improving, scale. Flat across the window, kill |
| Return rate | Whether returns are quietly eroding margin | Under 8% and steady, scale. Above it, fix the product |
| Conversion rate | Whether the traffic you buy turns into orders | Recovering after a listing fix, scale. Flat, kill |
| Cost of customer acquisition trajectory | Whether each new order costs less than the last | Improving, scale. Worsening across the window, kill |
Flapen figures as of September 2026. The window is 60 to 90 days, written down before the review.
Fix sits between scale and kill, and it carries the most work of the three. Diagnose before spending another dollar across listing quality, primary image click-through rate, conversion rate, ad performance, traffic channel activation, pricing, and return rate. Then read the four signals again in a fresh window.
The 50 operators here carry about 70 brands between them, and the majority of those brands reach profitability in their first year. Knowing when to stop is as important as knowing how to launch.
How many items a small catalog should carry
Phase 1 tests up to 4 products at the same time, and each one enters on 200 units. Four items at once is 800 units and $20,000 to $40,000 before a single one is proven. At $5K to $30K a month the honest number is one or two, because budget is the constraint rather than ambition.
Management here is priced per product, so catalog size shows on the invoice every month. One product costs $800, two cost $1,150, and three cost $1,500, with all 50+ services included and no commission.
What Amazon allows you to send, how a unit is prepared, and what comes out of each order are account questions. Confirm each one inside your own account before an item joins the plan.
What most agencies will not tell you about the items you carry
A monthly retainer earns the same whether an item thrives or limps, so nobody across the table volunteers the kill. Score the provider the way you score an item.
| What to score them on | What a pass looks like | Weight |
|---|---|---|
| The stop rule | Four named signals and a written window, agreed before the first order | Must-have |
| Who applies it | The named person on your account, reading your own numbers weekly | Must-have |
| What they would stop today | One item in your catalog, named on the call, with the reason | Must-have |
| What stopping costs them | Their fee falls when the item goes, or the incentive runs the wrong way | Must-have |
| What they got wrong before | One item they killed too late, and what the delay cost the client | Strong preference |
Flapen figures as of September 2026. You set the pass mark, in one written sentence, before the call.
Run those rows on us first. Cutting an item on a three-product account moves it to the two-product tier, so our own invoice falls from $1,500 to $1,150 a month. The contract runs month to month on 30 days' notice, and on exit you keep the Seller Central account, the campaigns, and the creative.
If nobody you interview can name what would make them tell you to stop, run the catalog yourself. That test includes us.
Related answers
- Amazon FBA seller
- Remote fulfillment with FBA
- Rank top Amazon brand management firms worldwide
- Top Amazon account management companies ranked
- Done-for-you Amazon management: the complete guide
One free thing to do this week, for the seller carrying one to three products at $5K to $30K a month. Open your own account and pull the last 90 days for your weakest item. Write four numbers by hand: rating trend, return rate, conversion rate, and what each order cost you in ad spend.
Then write one sentence beside them, naming the date you stop if none of the four improves. It costs you an hour.
Have those four numbers read by an operator in a free written audit, back with prioritized fixes inside 48 hours at no charge, from Flapen.
