How to Create Viral Growth
If you’re reading this, you’re probably chasing the same holy grail that has consumed other founders for years: sustainable, exponential growth.
You’ve likely heard all the standard advice:
- “Just build a great product and users will come”
- “Add a referral program with cash incentives”
- “Copy what [unicorn company] did and you’ll see results”
After spending years leading growth at Uber and advising companies like Dropbox, TikTok, and dozens of startups, Andrew learned that most of this conventional wisdom isn’t just unhelpful, it’s dangerously wrong.
True viral growth doesn’t come from bolting on growth tactics to an existing product. It comes from deeply understanding how network effects work and engineering them into your product’s core experience.
The Network Effect Taxonomy: Which Type Are You Building?
The first mistake most founders make is assuming all viral growth works the same way. It doesn’t.
There are fundamentally different types of network effects, each requiring its own growth strategy:
- Direct Network Effects: Value increases directly with more users (Facebook, WhatsApp)
- Marketplace Network Effects: Value increases with more participants on both sides (Uber, Airbnb)
- Data Network Effects: Product improves as more usage data is collected (Google, Netflix)
- Compatibility Network Effects: Value rises as more complementary products join (iOS, Windows)
At his first startup, they blindly copied social sharing mechanics from Facebook, expecting similar results. Their viral coefficient never exceeded 0.3, meaning each user brought in less than one-third of another user.
Why? He says that they were a productivity tool with primarily compatibility network effects, but they were using growth tactics designed for direct network effects.
They were solving for the wrong equation.
Action Step: Map out exactly how additional users make your product more valuable. Then study the growth patterns of companies with the same network effect type. Don’t waste time implementing viral mechanics that don’t match your core value proposition.
The Cold Start Problem: Creating Value Before You Have a Network
Here’s the paradox every network-based product faces: your product isn’t valuable until it has users, but users won’t join until it’s valuable.
At Uber, this wasn’t just a theoretical problem, it was existential. Riders wouldn’t use the app if no cars were available, and drivers wouldn’t sign up if there were no riders requesting trips.
He explained:
Our solution was counterintuitive: extreme constraint. We focused exclusively on black cars in downtown San Francisco. This wasn’t a limitation, it was a deliberate strategy to create density in a small area where the network could actually function well.
We measured wait times obsessively in this concentrated zone, only expanding geographically once we hit our density targets. The same strategy worked city by city.
This applies differently based on your network type:
- For social products: Create a single-player utility that doesn’t require a network
- For marketplaces: Focus on constrained liquidity in a small geographic area or niche
- For data products: Pre-seed with existing datasets
- For platforms: Build first-party applications that demonstrate the value
Action Step: Define the smallest viable network that can deliver your core value proposition. For marketplaces, this often means a single neighborhood or vertical.
For social products, it might be a specific community or use case. Focus all resources on dominating this initial network before expanding.
The Viral Equation: Making Growth Measurable
True viral growth isn’t magic, it’s math. We can break it down into measurable components:
- i: Invites sent per user
- c: Conversion rate of those invites
- t: Time delay of the viral cycle
- 1/t: Frequency of the viral cycle
- k = i × c: Viral coefficient (must exceed 1 for exponential growth)
At Dropbox, Andrew found that offering storage to both the sender and recipient created perfect incentive alignment.
By measuring each component separately, they identified that the time delay (t) between receiving a referral and accepting it was their biggest opportunity. A simple follow-up email sequence reduced this delay by 40%, significantly accelerating growth.
At Uber, their referral program initially had a high invitation rate (i) but a poor conversion rate (c). People were sending mass invites to their entire contact list, resulting in low-quality referrals.
When they redesigned the program to encourage more targeted sharing with close contacts including the ability to share a specific ride experience, their conversion rate doubled, even though the raw number of invites decreased.
Action Step: Build analytics that separately track invitation rate (i), conversion rate (c), and cycle time (t).
Run experiments focused on improving your weakest metric first. Remember that your viral coefficient must exceed 1 for true exponential growth.
The Atomic Network Strategy: How Facebook Really Grew
One of the most misunderstood aspects of Facebook’s growth is how deliberately they expanded through what Andrew calls “atomic networks”—discrete social clusters where users already had connections.
Network effects don’t spread evenly across an entire user base but through existing social structures.
Facebook’s campus-by-campus expansion strategy wasn’t just about exclusivity, it was about ensuring complete penetration of each atomic network before moving to the next.
They would launch at a new college only when they had captured 70%+ of the previous campus. This ensured the network was valuable immediately upon launch, as most students already had friends using the platform.
Andrew has applied this same principle with marketplace startups. Rather than trying to cover an entire city with mediocre liquidity, they focus on dominating specific neighborhoods completely before expanding.
Action Step: Identify your minimum viable network unit (office, campus, neighborhood). Create density metrics to track penetration within each atomic network. Focus on dominating these atomic networks completely before expanding to adjacent ones.
Your Core Action Loop: The Heart of Viral Design
At the center of every truly viral product is what he calls the “core action loop”—the fundamental sequence that creates and distributes value through the network.
When analyzing TikTok’s extraordinary growth, he identified that their core loop wasn’t just about consumption but creation and sharing.
Each view could lead to a creation, which generated more views. The entire product was designed to make this specific sequence as frictionless as possible.
Pinterest’s core action loop was brilliantly simple:
pin something interesting → others discover and repin it → more people see your content → you’re motivated to pin more.
The mistake most founders make is focusing on peripheral features while neglecting to optimize this fundamental loop.
Action Step: Map out the 3-5 key steps in your product’s core interaction. Time how long each step takes and systematically eliminate friction points. Then identify how this core action naturally exposes your product to non-users and amplify those touchpoints.
The most important insight you should know is that sustainable viral growth isn’t about clever hacks, it’s about fundamentally engineering your product so that its core value proposition naturally creates network expansion.
The companies that win don’t just find temporary growth loops; they build self-reinforcing systems where each new user makes the product more valuable for everyone.