Network vs Ecosystem Effects
These get conflated often, but they’re very different.
Network effects: who else is here? each additional user makes the network more valuable to every other user. This can take two forms: direct (more users = more value, like a phone network or messaging app), or indirect (more users on one side attract more on the other, like a marketplace). The key property is that value scales with n, the number of participants in the same network. Metcalfe’s, Reed’s, etc. are all just different exponents on this core dynamic.
Ecosystem effects: what can you do here? are about breadth across complementary products or services. Value accrues there are more things you can do within a connected system. iPhone’s moat it’s that the density of apps, accessories, services, and integrations makes switching away from the whole stack prohibitively costly. AWS is similar: no network effect between customers, but the ecosystem of services, integrations, and tooling creates compounding lock-in.
It matters strategically too. Network effects tend to be winner-take-most within a category but are vulnerable to multi-tenanting (users can be on multiple networks simultaneously — think messaging apps). They’re strongest when the network is the product; e.g. LinkedIn nailed this.
Ecosystem effects are stickier against multi-tenanting (you can’t realistically use two ecosystems for the same workflow, e.g. Android and iOS). They defend through switching costs and workflow integration.
The most durable businesses layer both. E.g. Stripe has network effects on the payment processing side which improves their unit economics drastically (e-commerce payment processing is all about the unit economics of chargeback management): more merchants → more fraud data → better acceptance rates. And they have strong ecosystem effects (Billing, Atlas, Connect, Identity make it painful to leave).
At Peanut, this matters to us in two ways:
P2P builds network effects. That’s obvious. Think Venmo.
Ecosystem effects are more complex. Whoever we integrate (stablecoin, bridging solution, etc.) needs to have ecosystem effects in place already. That’s why we are e.g. uninterested in alt L1s or our own branded stablecoins unless they can prove they have the stack we need. (This might change as we grow. There is a function where as you reach enough scale you can start verticalising; Stripe did this, starting with a single API call and gradually absorbing more of the financial stack until leaving meant rebuilding half your backend. But that’s a luxury you earn at scale. Right now we’re better off borrowing ecosystem effects from partners who’ve already built them (Arbitrum’s liquidity depth and AA infra, Bridge’s fiat rails, USDC’s CCTP etc) rather than trying to cultivate our own. This flips when the cost of dependency exceeds the cost of building (not the software, the ecosystem!); and we’re nowhere near that threshold yet.)
The spend side is where ecosystem effects live for us. The card, local payment methods, bill pay, rent… every additional thing you can do with your Peanut balance makes the balance stickier. This is what we got wrong in Buenos Aires: we had decent network effects during DevConnect (people sending money to each other, referring friends, the viral loop doing its thing) but not much in terms of ecosystem effects outside of Argentina and Brazil. The moment people left Argentina and lost spend-side access, there was nothing anchoring them. The balance became useless and got moved out. The network effect was real but fragile because it wasn’t backed by an ecosystem that gave people reasons to keep money in Peanut.
Launching a card converts Peanut from a transfer network (which has network effects but weak retention) into a financial surface area (which has ecosystem effects and high switching costs).
The distinction maps onto something deeper: network effects are about who else is here, ecosystem effects are about what else you can do here. The first is social, the second is structural. Social is powerful but fickle: people follow people. Structural is slow but compounding: people follow workflows. We’re build the social layer to get people in. We’re build the structural layer to make leaving irrational.