The five models, and who each one is wrong for

Side by side

Ranges are compiled from published seller guides, not audited data. See what it costs to start for why the private label range is so wide.

Retail arbitrage

Buying clearance and markdown stock from physical stores and reselling it. Lowest cost of entry and the fastest route to a first sale, but it scales only as fast as you can physically walk shelves, and it builds no asset you could ever sell.

Online arbitrage

The same idea without leaving the house: sourcing discounted stock from online retailers. Better suited to people with time and patience for screen work than to people who like driving. Same ceiling as retail arbitrage, same lack of a durable asset.

Wholesale

Buying established branded products in bulk from authorised distributors and reselling them. More predictable than arbitrage and genuinely scalable, but it lives or dies on supplier relationships — which means cold outreach, account applications, and being taken seriously as a business.

Private label

Sourcing a generic product, branding it, and owning the listing. The only model on this list that builds something you could eventually sell. Also the slowest to first revenue, the most capital-hungry, and the one where a single bad product decision can consume your entire budget.

Dropshipping

Listing products you do not hold and having a supplier ship them. Lowest capital requirement and lowest margin. Amazon's drop shipping policy has strict conditions on who must appear as the seller of record on packing slips and invoices — read it before you build anything on this model.

What decides it, in practice

Four constraints do most of the work, and none of them is enthusiasm.

• Capital you can lose. Not capital you have. The models sort almost cleanly on this one axis.
• Supplier access. Wholesale is gatekept — a registered business and resale certificate are usually prerequisites for a distributor account. No amount of money substitutes for that paperwork.
• Payout timing tolerance. The gap between paying a supplier and being paid by the marketplace decides whether you can reorder your winner.
• Margin buffer. The 2026 fee changes hit thin-margin models hardest. Models running at five to ten percent net have almost nothing left to absorb an increase; models above fifteen percent can take it.

The one nobody says out loud

Arbitrage and dropshipping are frequently sold as beginner-friendly. They are cheap, which is not the same thing. They also carry the thinnest margins, the least protection against fee changes, and no asset at the end. They are a good way to learn how Seller Central works with money you can afford to lose. They are a poor destination.

Conversely, private label is sold as the sophisticated choice, which encourages undercapitalised people to attempt it. It is the model where being underfunded is most expensive, because the failure mode is a container of unsold branded inventory that nobody else can sell for you.