Why peer-to-peer marketplaces die at small ticket sizes

When you sell a $20 thing on eBay, you keep about $17.

The other $3 goes to two places stapled together: roughly $2.95 to eBay (a 13.25% final value fee plus a $0.30 fixed costeBay's standard final value fee is 13.25% across most categories as of 2025. A few categories run lower (Books, DVDs & Movies are 14.95% on the first $7.50 then 2.35%; some collectibles tiers exist). The $0.30 fixed cost applies per order regardless.), and the rest to payment processing buried inside that fee. The exact split varies, but the shape doesn't. About 15% of every small trade evaporates before it reaches the seller.

That 15% is not arbitrary. It's the price of two things bundled into one number: the cost of moving money on card rails, and the cost of running the trust layer (buyer protection, dispute resolution, fraud underwriting) that makes the trade safe in the first place.

The interesting question is which of those two is the bigger number, and what happens to peer-to-peer commerce when one of them goes to zero.

The trust layer is mostly cheap

Running escrow, in software terms, is not expensive. You hold funds. You watch for delivery confirmation. You run a dispute resolution process for the small fraction of trades where someone complains. Amortized across all transactions, the per-trade cost of running the trust layer is in the tens of cents. Maybe a dollar at the top end if you do detailed identity verification.

eBay has thousands of dispute-resolution employees, sure. But eBay processes billions of transactions a year. The headcount cost per trade is small.

The rails are most of the fee

What's expensive is moving the money. Card networks charge interchange (Visa and Mastercard's slice, ~1.5-2.5%), plus the processor's cut (Stripe / Square / Adyen, ~0.3-0.5%), plus chargeback risk underwriting, plus the cost of FX where applicable, plus the ACH or wire fees on the seller payout side.

Add it up and a typical card-rail trade burns 3-4% on payment processing alone before any platform fee is added.

So when eBay charges you 13.25%, roughly a third of that is paying for the rails (which they have to in turn pay Visa and Mastercard and Stripe), and the rest is platform margin, dispute infrastructure, and operating costs.

Why this kills small-ticket P2P

A $20 trade with 15% rake gets the seller $17 and the platform $3. Livable, but barely.

A $5 trade with 15% rake gets the seller $4.25. Now the seller probably doesn't bother.

Below about $10, the fee structure stops working. Sellers list elsewhere or pocket cash directly, the platform doesn't see the volume, and a whole category of trade just doesn't happen at scale through formal marketplaces.

This is why eBay fights you at small sizes. It's why Etsy keeps adding fees. It's why Mercari pivoted toward higher-AOV categories. Card rails impose a floor on what's economically viable, and that floor is somewhere in the high single digits.

The thought experiment

Strip out the rails. Don't change anything else. Same dispute resolution, same buyer protection, same software, same operating costs. Just the rails are zero.

What does the seller's net look like on a $20 trade?

If you charge a flat fee that reflects the actual cost of running the platform (let's say 4%), the seller keeps $19.20. The platform takes $0.80. That's a real business at scale, and it works at $5 trades too.

The only thing that changed is the cost of moving the money. The trust layer is the same. The economics on small trades become possible because the variable cost on the rails is now zero.

Why Nano

Bitcoin doesn't work for this. Fees are unpredictable and frequently above $1Bitcoin median transaction fees over the last two years have ranged from roughly $0.50 to over $30 during congestion spikes. Even the floor kills small-ticket commerce; the ceiling kills it spectacularly., which kills the floor immediately.

Ethereum L1 is worse. L2s are improving but aren't there yet for arbitrary value sizes.

Lightning is close on the technical side, but liquidity routing remains a real problem for arbitrary peer-to-peer tradesA Lightning payment between two arbitrary parties needs a routable path of channels with sufficient inbound and outbound liquidity at every hop. For predictable sender-recipient pairs (a customer paying a coffee shop) this is a solved problem. For a marketplace where any user might pay any other user, it is not., and final settlement back to the base layer eventually costs something.

Nano is the only network I'm aware of where settlement is genuinely feeless, instant, and final. It uses a directed acyclic graph instead of a chain, with proof-of-work rate-limiting instead of fee-based rate-limiting.Each Nano account has its own block-lattice; senders compute a small proof-of-work to attach a transaction. There is no miner to pay because confirmation is reached by representative voting weighted by delegated stake, not by burning hashpower. See nano.org for the full design. There is no fee, ever, by design. Settlement is in the second-or-two range. There is no chargeback because there is no card network behind it.

This isn't an endorsement of Nano as a speculative asset. The price moves around like any small-cap crypto. The argument is narrower than that: as a payment rail, the network has the properties a peer-to-peer marketplace needs, and Bitcoin doesn't.

What I built

I spent the last year building Nagora, a peer-to-peer marketplace settled on Nano. Sellers pay a 4% flat fee with no card processing on top. Buyers pay nothing to the platform. Every trade goes through escrow by default, not as an opt-in. There's no KYC at launch, just an email and password.

I won't pretend the day-one catalog is impressive. I seeded it with Amazon and Steam gift cards across common denominations because an empty marketplace is a dead marketplace. The actual goal is for other people to list things, and there are now a handful of non-gift-card listings ⊕As of late April 2026: a used GPU, some shoes, a Bitcoin Ordinal, a few promotional Steam keys. Browse the live catalog → which I'm cautiously encouraged by.

The first 15 verified sellers this month pay 0% forever and get a permanent founder badge on their profile. Six of those slots are still open.

What I'm watching

The honest caveats. The catalog is small. I'm a solo developer. No KYC at launch is going to come up as a regulatory question and I'll be upfront if that changes. The in-app wallet is for funding and receiving trades, not for holding balances; the right pattern is fund, trade, withdraw.

One thing card rails do that Nano doesn't: lock the fiat-equivalent for both sides at point of sale. Right now I take the buyer's XNO at checkout and hold it in escrow at that quantity. If XNO moves materially during the escrow window (typically hours, sometimes days), the seller's eventual fiat conversion drifts. I haven't built a hedge or a price-lock yet. The plan is to give trusted sellers an opt-out from escrow entirely once they've built reputation, which compresses the volatility window to seconds.

The argument I want to test isn't really “Nano specifically.” It's that feeless final settlement enables a category of commerce that card rails make uneconomical. Nano is the first network where this is true today. Lightning is approaching it. Other feeless networks may emerge. The economics generalize.

If the experiment works, it's because the floor on viable trade size dropped from $10 to roughly $1. If it doesn't work, it's probably because peer-to-peer commerce has structural problems beyond fees (counterparty trust, fulfillment logistics, category fragmentation) that lower fees don't fix.

I think it's worth testing. If you've spent any time thinking about why small-payment commerce never quite worked online, I'd be interested to hear where you'd push back.