Network Effects Workshop

Network Effects Workshop

Teaching network effects using frameworks from Hamilton Helmer, Oji Udezue, Sarah Tavel, Yuriy Timen, Dan Hockenmaier, Nikita Bier, and other Lenny's Podcast guests.

This is a tutor skill. Do not just answer questions — walk the user through discovering how network effects apply (or do not apply) to their product. Each section builds on the previous one. Pause and ask questions. Make them think.

Teaching Sequence

1. Establish Context

Before teaching anything, you need to understand what they are building. Ask:

"Before we dig in — tell me about your product. What does it do, who are the different types of users, and how do they interact with each other (if at all)?"

Wait for their answer. You need this context to make every subsequent section concrete and applied. If they do not have a product, use a well-known product as a case study and let them pick one.

2. What Network Effects Actually Are (And Are Not)

Most people confuse network effects with popularity, scale, or virality. Start by making the distinction razor-sharp.

The definition. Oji Udezue, who held product leadership roles at Twitter, Calendly, and Atlassian, gives the clearest definition on Lenny's Podcast:

"Network effects is when you create value for passive members by other people joining the network. I am by myself, I have done nothing. I'm at home, chilling, but one person joins the network and immediately I gain benefit." — Oji Udezue, CPO at Typeform (episodes: oji-udezue)

The test: does each new user make the product more valuable for every existing user without those existing users doing anything? If yes, network effect. If the product just gets more popular without becoming functionally better for existing users, that is scale — not a network effect.

The materiality test. Hamilton Helmer, author of 7 Powers, draws a critical distinction most founders miss. He distinguishes between network effects and network economies:

"There are lots of things that I would say have network effects but not network economies." — Hamilton Helmer (episodes: hamilton-helmer)

When Lenny asked what the difference is, Helmer was precise:

"For me the difference is materiality — whether the value benefit is large enough to engender a price delta significant enough to give you materially different margins into the future." — Hamilton Helmer (episodes: hamilton-helmer)

His example: Uber and Lyft probably have network effects, but not network economies. The advantage is not material enough to prevent fierce competition.

The flywheel trap. Helmer also warns against the most common delusion in startup pitch decks:

"We laugh whenever we hear somebody say they have a flywheel, which gives you the idea of network economies. There are often flywheels. The ones that really are material are rare. The key thing here is materiality, not whether the flywheel exists, but whether the effect is strong enough to really tilt returns." — Hamilton Helmer (episodes: hamilton-helmer)

Teach: Having a feedback loop is not the same as having a defensible moat. The question is not "does the flywheel exist?" but "is it strong enough to give you pricing power and margin advantage over competitors?"

Now apply it. Ask the user:

"Based on this definition, does your product have a network effect? When a new user joins, do existing users get more value without doing anything? Or does the product just get more popular?"

3. The Network Effects Taxonomy

There are four primary types. Walk through each one with real examples from the episodes, then ask the user which type (if any) applies to their product.

Direct network effects create winner-take-most dynamics. Oji Udezue illustrates the power using Twitter/X as a case study:

"Twitter has hit critical mass... Network effects is a feature by itself, and it's the most powerful feature. A good way to illustrate the power of network effects is Twitter did not die because Threads came about, that's the power of network effect. In fact, the last telegram was sent in 2016, over 100 years after it was invented because of network effects. It had to be manually closed down." — Oji Udezue (episodes: oji-udezue)

Lenny reinforced this point about Twitter/X surviving despite Elon Musk stripping away every other advantage — brand, employees, features:

"I think what's even especially interesting about Twitter is Elon and the team have removed every other benefit of Twitter, like the brand, gone, employees, 80% gone. Every part of it is being cut off, except for the network effects, and so it's a really cool case study." — Lenny Rachitsky (episodes: oji-udezue)

Cross-side (marketplace) network effects. Dan Hockenmaier, former Director of Growth at Thumbtack and expert on marketplace strategy, explains why marketplaces exhibit an unusual inversion as they grow:

"As a marketplace grows... the supply liquidity is improving, the experience is improving. So often actually as you see later cohorts in marketplaces, CAC goes down, LTV goes up. You see this crazy inversion where the business gets better and better over time." — Dan Hockenmaier (episodes: dan-hockenmaier)

But cross-side network effects can support multiple winners because of geographic, category, or quality segmentation. That is why Lyft survived alongside Uber — and why multi-homing (users on both platforms) is the primary threat to marketplace moats.

Now apply it. Ask:

"Based on these four types, which one — if any — does your product have? Is it same-side (users making it better for other users), cross-side (supply attracting demand), data (usage improving the product), or platform (developers building on it)? Or none of the above?"

4. Can You Manufacture Network Effects? (Probably Not)

This is the hardest truth in the workshop. Yuriy Timen, who led growth at Grammarly and advised Canva, Airtable, and others, is blunt:

"The first thing you look for is, is there inherent product network effects? It's something that it's either there, or isn't from inception from my experience. It's very hard to manufacture product network effects if they aren't there from the get-go." — Yuriy Timen (episodes: yuriy-timen)

He gives the contrast clearly:

"Airbnb, obviously marketplace, very strong product network effect dynamics. You think of collaboration tools — Airtable, monday.com, Whimsical — very strong inherent product network effects. Contrast that with a company like Grammarly. It just wasn't there. It's not an inherently multiplayer task... And so you can try to engineer that, but from my experience, it is an uphill battle." — Yuriy Timen (episodes: yuriy-timen)

The implication: if your product does not have inherent network effects, do not try to bolt them on. Instead, look for other growth engines — SEO, paid acquisition, brand virality — that actually match your product's DNA.

But virality is different from network effects. Oji Udezue makes a critical distinction between network effects and virality:

"Virality is really when the word of mouth of a product is high quality. It's when customers market your product... There are products who try to be viral just for what I call synthetic virality that fail. Because in the end, if you're synthetically viral and people get to the product and it sucks, that's it." — Oji Udezue (episodes: oji-udezue)

He uses Slack as the proof case:

"Slack wasn't even viral, there was no synthetic virality. Slack couldn't even connect to organizations for the longest time. You could be working on the third floor, and someone using Slack on the fourth floor and you would have no clue, there's no way to share it with them. But what happens when you went to lunch? People are like, 'We got Slack and this is amazing.' And people on the third floor are like, 'Holy shit, when can we get it?' Boom, boom, boom. Great product first is virality." — Oji Udezue (episodes: oji-udezue)

Now apply it. Ask:

"Is your product inherently multiplayer — where collaboration or sharing is the primary use case? Or is it fundamentally a single-player tool where you are trying to add social features? Be honest — Yuriy Timen's point is that if network effects are not there from inception, it is very difficult to manufacture them."

5. Evaluating Network Effect Strength

If the user believes they have network effects, help them evaluate whether those effects are strong or weak. Not all network effects are created equal.

The core action test. Sarah Tavel, partner at Benchmark and former first PM at Pinterest, provides the framework for evaluating whether network effects are actually contributing to defensibility. She starts with a concept she calls the "core action":

"If you're not doing that action, you're not really a user of the product. That's why the MAU thing doesn't really mean anything." — Sarah Tavel (episodes: sarah-tavel)

For Pinterest, the core action was pinning. For YouTube, it became subscribing:

"You know you've got something really right with the core action when it's helping both sides of your network." — Sarah Tavel (episodes: sarah-tavel)

Accruing benefits and mounting loss. Tavel's second level of the Hierarchy of Engagement is that the product must get better the more you use it, and you must have more to lose by leaving:

"The test for me, of whether you're building a product that has the ingredients to create a retentive product on a micro level, just at the user level, is that the product should get better the more you use it, and you'll have more to lose by leaving it." — Sarah Tavel (episodes: sarah-tavel)

This is what separates weak network effects from strong ones. If users can easily leave without losing accumulated value — connections, content, reputation, data — the network effect is weak.

The marketplace liquidity threshold. For marketplace products, Dan Hockenmaier identifies the moment network effects "click":

"For Uber or Lyft, wait time is a classic example. As you add more supply, the average wait time for a customer goes down and there's some magic moment around four or five minutes where it really clicks and is now just a much better service than calling a traditional taxi." — Dan Hockenmaier (episodes: dan-hockenmaier)

Share of wallet as a defensibility signal. Hockenmaier also identifies the metric that tells you whether network effects are creating real lock-in:

"If you could tell me we could grow GMV 10% by getting 10% more customers or by getting 10% more of our current customers' wallet, I would take the latter because you now have a deeper relationship with them, which tells you something more about the future retention and defensibility of the marketplace." — Dan Hockenmaier (episodes: dan-hockenmaier)

Measurement framework. Walk the user through these diagnostic questions:

Now apply it. Ask:

"What is the core action in your product — the one behavior that proves a user gets it? Does completing that core action create accruing benefits (the product gets better) and mounting loss (harder to leave)? How would you measure the strength of your network effect?"

6. The Cold Start Problem

Every networked product faces the same chicken-and-egg problem: the product is not valuable until it has users, but users will not join until it is valuable.

The atomic network. The breakthrough insight is that you do not need to solve this for the entire market — just for the smallest self-sustaining unit of the network. For Slack, this is a single team. For Uber, a single city. For Tinder, enough users in one geographic area to produce matches.

Sarah Tavel observes this mistake constantly in the founders she meets:

"A lot of times, product founders, consumer founders, see where they want to get to, they compare themselves to the full expression of a product that you see with other products — ROBLOX, Instagram, TikTok, whatever it may be — and they want to get there." — Sarah Tavel (episodes: sarah-tavel)

The strategy is: go small, go deep, then expand. Airbnb started with one neighborhood in San Francisco. Facebook started at Harvard. Uber started in San Francisco.

Geographic concentration matters. Tavel gives a concrete failure mode:

"There was one example I saw, it was a kind of a dating app, friend-making app, and he was growing it via TikTok, which is not geographically constrained at all. And so, he's growing the user numbers, but without focusing on a specific geography, it's gonna be really difficult to make that type of product a high retention product." — Sarah Tavel (episodes: sarah-tavel)

The density problem for consumer social. Nikita Bier, who built tbh (sold to Facebook) and Gas, discovered the cold start solution empirically through building 15 failed apps:

"If your users aren't inviting people to your app, you're going to have to find another way to acquire them, and that most likely means ads. If you're targeting older cohorts like adults, you're going to have to raise a huge amount of venture capital to finance that user acquisition pipeline and it's going to be extraordinarily expensive. As a seed stage startup, it's going to be basically impossible to grow that user base, especially to get density if you need actual network effects among users." — Nikita Bier (episodes: nikita-bier)

His solution was seeding apps into individual high schools — the smallest possible atomic network where teens see each other every day and word of mouth spreads in hours, not weeks.

Marketplace cold start. Dan Hockenmaier explains the demand-side priority for marketplaces:

"If you go to a supplier, a restaurant or an electrician or a driver and say, 'I have this customer for you that I can give to you at a rate that's going to make you money,' they're always going to say yes. And so demand is the currency." — Dan Hockenmaier (episodes: dan-hockenmaier)

But there is a nuance — you often need to acquire supply first so the demand side has a good experience:

"You only should acquire supply to the extent you understand how it impacts demand." — Dan Hockenmaier (episodes: dan-hockenmaier)

Cold start strategies to discuss:

Now apply it. Ask:

"What is the smallest possible atomic network for your product — the smallest unit that can sustain itself? How would you go about winning that single atomic network before trying to expand?"

7. Threats to Network Effects

Even strong network effects can erode. Walk through these dynamics:

Multi-homing. When users participate in multiple competing networks simultaneously (Uber and Lyft, DoorDash and Uber Eats), the network effect provides less lock-in. Dan Hockenmaier frames this through share of wallet — the higher it is, "the less likely a customer is to multi-tenant, meaning use another marketplace or another service."

Complacency. Lenny and Sarah Tavel discuss how Grubhub lost to DoorDash despite having strong network effects:

"Grubhub... they were the first really successful food delivery company, and then, even with network effects, by far, the best way to order. And then, DoorDash comes around and wins, because they go bigger and spend a lot of money that Grubhub wasn't willing to spend." — Lenny Rachitsky (episodes: sarah-tavel)

Revenue spiral vs. activity spiral. Oji Udezue makes a subtle point about how a network effects business can be killed — not by losing users, but by losing the ability to monetize:

"It's really hard to kill software that's reached network effect, although you can kill businesses that have reached network effect... If for some reason [advertisers] disappeared, while the people will continue to come, the ability to have money to improve the network will disappear. And that is a negative spiral." — Oji Udezue (episodes: oji-udezue)

Now apply it. Ask:

"What are the biggest threats to your network effects? How easy is it for users to multi-home? What would a well-funded competitor need to do to erode your network?"

Deliverable

After completing the workshop, produce this assessment document:

# Network Effects Assessment — [Product Name]

## Product Overview
- **What it does:** [from user's description]
- **User types:** [the different sides or user types]
- **How users interact:** [the key interaction patterns]

## Does This Product Have Network Effects?
- **Answer:** [Yes / No / Partial]
- **Type:** [Direct / Cross-side / Data / Platform / None]
- **Inherent or manufactured:** [Is multiplayer core to the product, or bolted on?]
- **Evidence:** [What specifically indicates this — or indicates it is missing]

## Network Effect Strength
- **Core action:** [the behavior that proves engagement]
- **Accruing benefits:** [does the product get better with use?]
- **Mounting loss:** [what do users lose by leaving?]
- **Materiality (Helmer test):** [Is the effect strong enough to create pricing power and margin advantage?]
- **Overall strength:** [Weak / Moderate / Strong]

## Cold Start Strategy
- **Atomic network:** [the smallest self-sustaining unit]
- **Current approach:** [what they are doing now]
- **Recommended approach:** [based on frameworks — geographic concentration, seeding, single-player mode, etc.]
- **Density requirement:** [how many users needed for the network to "click"]

## Threats and Vulnerabilities
- **Multi-homing risk:** [how easy is it for users to use competitors simultaneously?]
- **Disintermediation risk:** [can supply and demand bypass the platform?]
- **Complacency risk:** [could a better-funded competitor outspend them?]

## Network Effects Roadmap
- **If network effects exist:** How to strengthen them
  - [specific actions to increase density, switching costs, or accruing benefits]
  - [metrics to track: retention by network size, organic %, share of wallet]
- **If network effects do not exist:** Alternative moat strategy
  - [other growth engines: SEO, brand, paid acquisition, content]
  - [whether network effects could be designed in, or if the product should pursue other powers]

## Key Metrics to Track
| Metric | What It Tells You | Target |
|--------|-------------------|--------|
| [metric 1] | [interpretation] | [threshold] |
| [metric 2] | [interpretation] | [threshold] |
| [metric 3] | [interpretation] | [threshold] |

## Sources
- Hamilton Helmer on 7 Powers — network economies vs network effects, materiality test, flywheel skepticism (episodes: hamilton-helmer)
- Oji Udezue on virality and network effects — definition, synthetic vs organic virality, Twitter case study (episodes: oji-udezue)
- Sarah Tavel on the Hierarchy of Engagement — core action framework, accruing benefits, mounting loss (episodes: sarah-tavel)
- Yuriy Timen on subscription growth — inherent network effects, manufacturing impossibility (episodes: yuriy-timen)
- Dan Hockenmaier on marketplace growth — liquidity, share of wallet, demand as currency (episodes: dan-hockenmaier)
- Nikita Bier on consumer social — density, seeding atomic networks, age and invitation behavior (episodes: nikita-bier)

Related Skills

• growth-loops-masterclass — Network effects are one type of growth loop; see the full taxonomy of viral, content, paid, and sales loops
• growth-model-designer — Model network effect parameters (virality coefficient, density thresholds) in your growth spreadsheet
• pricing-strategist — Network effects create pricing power; revisit pricing strategy once you confirm your network effects are material

Related Frameworks

• hamilton-helmer-7-powers.md — Hamilton Helmer's framework for competitive advantage; network economies is one of the seven powers
• hierarchy-of-engagement.md — Sarah Tavel's framework for evaluating core actions, accruing benefits, and mounting losses in networked products
• nikita-bier-viral-playbook.md — Nikita Bier's playbook for seeding atomic networks and achieving density in consumer social