Billion-Dollar Apps Don't Cold Start with Ads—They Start by Finding People
Deep thoughts on AI and aspirations —— ByteDance Deep Thinking Circle
A repeatedly misunderstood question: How do early-stage products get their first 1000 users? Many people’s intuition is “run ads, buy traffic, go viral.”
But looking at products that later became major companies, their cold start approach was almost never paid acquisition. Instead, it was something more fundamental—finding the right people and turning them into your product’s first believers.
The First 1000 Are Believers, Not Traffic
There’s a big difference between traffic and believers. Traffic is “people who’ve seen you,” believers are “people willing to speak for you.”
A few real cases make this clear. One social app’s earliest users came from the founder having one-on-one conversations on college campuses. One food delivery app’s first version was just a website and a stack of flyers posted around campus to test if anyone would order. One music app relied on invitation codes in the early days, making it “invite-only.”
What these approaches share: none were buying traffic. They were finding people who would genuinely love the product and be willing to talk about it. These people don’t need to be convinced with money—they’re the product’s most authentic early users and its best free advocates.
Why “Direct Contact” Is the Most Effective Cold Start
A counterintuitive observation: Nearly all successful products used “unscalable” manual methods early on—personally finding users, solving problems hands-on, one-on-one service.
Why does this work? Because early-stage products’ value is often vague and needs to be experienced firsthand. You can’t use an ad slogan to make everyone understand your value. You can only show it hands-on to some people, let them feel it, then they spread it for you.
A practical test: If you can’t personally help the first 100 users feel your value, you also can’t use ads to help the first 10,000 feel it. Cold start isn’t a marketing problem—it’s “finding people who can feel the value.”
Let Your First Users Speak for You
Another key action: Turn your first users into advocates.
Many early products deliberately give users a sense of “I participated in building this product”—through invite systems, communities, special status for early users. This isn’t about appearing exclusive—it’s about creating a sense of ownership: “this is part of me.” Users with a sense of belonging will proactively recommend you to friends.
This is the core loop of cold start: Find people who can feel the value → Turn them into believers → Let them speak for you. The first 1000 believers are the product’s best marketing team.
Two Truths You Must See Clearly
Don’t let these cases go to your head. Two things to stay clear about.
First, survivorship bias. You see successful products’ cold starts, not failed products’ cold starts. The same manual methods might succeed or fail. Cold start has no universal formula, only direction—find the right people, show them hands-on.
Second, the first 1000 and the first 10,000 require different playbooks. Cold start relies on manual methods (hands-on, one-on-one, believers). Scaling relies on something else (ads, channels, systems). Confusing these two stages is a common mistake: Some use cold start methods to chase volume and end up inefficient. Others start with ads right away, but the product value isn’t validated yet—money burned with no response.
Taking Action
The three priorities for cold start:
One, clearly define who your target users are, where they are today, how they live, what attracts them.
Two, personally find and serve the earliest batch of people (even if it’s just 10), let them genuinely feel the value.
Three, give this batch a sense of “I participated in the product,” make them willing to speak for you.
The first 1000 come from finding people, not buying traffic. This sounds simple, but many products never do it in their entire lifetime.
Key points: The first 1000 are believers, not traffic; direct contact is the most effective cold start; let early users speak for you; beware of survivorship bias, the first 1000 and 10,000 are different playbooks.