Network Effect

The Organisation for Economic Co-operation and Development (OECD) calls the influence of users on others in the same group a direct network effect. You have 10 friends on your messaging app; someone else you need to contact joins, and the service gains another useful possibility. A million strangers or inactive accounts might not help you in the same way. A reachable contact matters more than a number in an advertisement. Not every new account brings the same benefit to every user. A family chat, a work team and a public community may each need a different circle of people. [1]

An indirect network effect works through another group. OECD describes buyers and sellers on a platform as an example. Our marketplace has 100 buyers and 10 sellers. If it grows to 200 buyers while keeping 10 sellers, merchants gain a larger potential audience. Buyers do not necessarily gain a wider choice of goods. More suitable sellers might subsequently bring that choice. A similar relationship can develop between a system’s users and developers of useful add-ons. In payments, Merchant Adoption means merchants starting to accept a payment method. Count buyers, places to buy and add-ons separately, and ask what each group actually gains. [1]

A large business can have lower costs even when its customers do not affect one another. OECD distinguishes this from network effects. Consider a workshop with total costs C(q) = 1 000 + 2q: q is the quantity produced, the fixed expense is paid regardless of output, and the remaining part rises with production. At q = 100 the average is 12 per item; at q = 200 it is 7. The same fixed expense has been spread over more items. That alone does not establish a Network Effect: your neighbour buying a product need not make yours more useful. Nor must the manufacturer pass every saving on to your price. [2]

Additional participants do not necessarily improve everyone’s experience. OECD discusses choice overload, privacy concerns and other limits to large networks. A new marketplace seller adds choice for buyers but competition for existing sellers. In a group chat, more messages can make important information harder to find. Improvement for one side is therefore not proof of improvement for all. You need to specify whose experience you mean and what actually changed. An account count without information about quality and service capacity cannot answer that. [2]

If your friends use different apps, you need not choose just one. OECD calls simultaneous use multi-homing. In our example, network A has 100 people, B has 80, and 60 use both. Together that is 100 + 80 − 60 = 120 distinct people, not 180: simple addition counted the shared users twice. One service’s growth also need not mean departures from another. Distinguish data portability from interoperability too. Taking your contact list with you makes moving easier, but does not by itself enable ongoing communication between two apps. [3]

Connecting people does not require a single company to own everything. The technical standard RFC 6120 describes XMPP, a protocol under which different servers can exchange messages. Such interconnection is called federation. Section 13.14 makes federation optional: a provider need not enable it. Two services displaying XMPP therefore do not automatically promise that their users can communicate. Actual connections, supported features and availability matter. Other systems likewise need to be assessed under their own rules. Nostr, for example, is an open protocol for social applications. An open standard enables a possible connection; it does not confirm that everyone can reach everyone. [4]

CFA Institute, a professional body for investment practitioners, discusses models based on network connections and their limitations. Imagine separate groups of 10 people. If there are 3 such groups, each fully connected internally, each has 45 possible pairs, or 135 in total. Connecting all 30 people produces 30 × 29/2 = 435 pairs: everyone has the others as possible counterparts, and division removes double counting. No people were added; the arrangement changed. Yet neither 135 nor 435 is an observed trade count or a monetary network value. Market liquidity—the ability to exchange the required amount without substantially worsening the price—also requires actual offers; a contact count cannot guarantee it. [5]

OECD describes incentives to attract participants and the conditions behind market dominance. Our campaign brings 1 000 registrations, but 200 members of that group return in the specified following month. That means 20% of the group used the service again. This alone does not measure a network effect: it records repeat use, not why people came back. Look at usefulness after rewards end, whether other participants help users, and how readily they can move elsewhere. For money, Monetary Network Effect explores acceptance and usefulness in exchange. Network Effect generally guarantees neither monopoly nor an increase in bitcoin or share prices. An advantage needs explaining through concrete use. [1]