Learn it → Apply it → Save as Skill
A product gets more valuable to its existing users as more users join. That is what separates a real network effect from ordinary growth — and it is why some market leaders become almost impossible to dislodge, while others with huge user counts turn out to have no real moat at all.
The Dr. Wallace’s Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
When each new user makes a product better for the users already there — more matches, more content, more people to reach — that is a network effect. When new users just add revenue without improving the product for anyone else, that is ordinary growth, not a network effect.
Important: this shows how fast possible connections multiply, not proof of value. Most real networks fall short of this because not every connection is equally useful — see “History & Accuracy” below.
The model is most useful when evaluating platforms, marketplaces and communities where growth is supposed to create a moat.
Network-effect analysis separates real defensibility from raw user counts.
Does using the product make it better for the users already there, or does it just add one more paying customer?
Direct (one group, like a messaging app) or indirect / two-sided (two groups that need each other, like riders and drivers)?
Track engagement or liquidity per side, not total signups. A platform full of dormant accounts shows weak effects despite a big headline number.
Check multi-homing and locality before calling the advantage a moat. Total scale alone is not the same thing as a network effect.
Open a use case to see what the model changes.
Before subsidizing user acquisition to “win” a market, check whether the product actually gets more valuable per user as it scales, or whether the company is simply buying market share that a better-funded rival could copy at will.
A personal to-do list does not get better because other people use it — that is a personal-utility feature, not a network effect. Adding “share this list with a team” is what would actually introduce one.
Use this model to judge whether a large user base is a durable competitive advantage or a number that could evaporate if users can easily switch or use a rival at the same time.
Decide which side to subsidize first and where to reach critical mass — usually within one city, category or niche rather than nationally — before expanding.
Ask whether your existing user base creates real defensibility against a challenger with more capital, or whether users could and would multi-home or switch if offered a simultaneously better deal.
Internal platforms such as a company wiki or chat tool show network effects at the team level. Consider reaching critical mass within one team before rolling a tool out company-wide.
This is an illustrative example designed to teach the method.
Expand for the distinction between the popular formula and the academic economics behind it.
Robert Metcalfe, a co-inventor of Ethernet, argued in the 1980s that a network’s usefulness grows with roughly the square of its connected users, since the number of possible connections between n users is n(n−1)/2. It began as a sales argument for buying more networking equipment, not as a peer-reviewed economic result.
In 1985, economists Michael Katz and Carl Shapiro published foundational work on “network externalities,” giving the underlying phenomenon rigorous academic treatment as a demand-side economy of scale — years before “network effects” became common startup vocabulary.
Critics, including internet economist Andrew Odlyzko, have argued that Metcalfe’s n² substantially overstates real network value, since not every connection matters equally — most people care about a small circle of contacts, not everyone else on the network. More moderate growth models have been proposed. Treat Metcalfe’s Law as directional intuition about accelerating value, not a precise valuation formula.
Viral growth is about efficient acquisition — referrals, sharing, PR. A network effect is about the product getting better as users join. A product can grow virally with no network effect at all.
More users can lower unit costs through bulk purchasing or amortized fixed costs without making the product itself better for existing users. That is a scale advantage, not a network effect.
Users active on several rival platforms at once — like drivers on two ride-hailing apps — weaken the lock-in a network effect is supposed to provide.
Dormant accounts and one-time sign-ups create no value for anyone. Active, engaged usage on the relevant side is what matters.
Some networks get worse past a point: crowded roads, noisy feeds, oversaturated marketplaces. Growth is not always good.
Effects usually only appear after a network clears local critical mass. Before that, more users can mean more chaos, not more value — the chicken-and-egg problem.
the product’s value comes mostly from individual utility that does not change with other users, such as a personal note-taking app or a calculator.
the real advantage is bulk purchasing power, amortized fixed costs, or brand recognition rather than the product improving for each user as more join.
the “lock-in” comes from data-portability friction, contracts, or integration costs rather than other users’ presence making the product better.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the Network Effects model, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
No. A large user base only becomes a moat when existing users would lose real value by leaving — because the product gets worse without the network, not merely because many people happen to use it today.
A two-sided network is worthless to side A until side B is present, and worthless to side B until side A is present. Common solutions include subsidizing one side first, seeding supply manually, launching in a single dense niche or city, or building a single-player mode that has value before any network exists.
Yes. Past a point, more users can crowd a road network, fill a social feed with noise, or flood a marketplace with low-quality listings. A network effect is a direction of change in value, not a guarantee that the direction stays positive forever.