Nikita Bier on Viral Consumer Apps, Latent Demand, and Building for Teens

Nikita Bier has launched more apps to the top of the US App Store than any other individual. He sold tbh to Facebook for over $30 million (nine weeks after launch) and Gas to Discord for additional millions. He built both with tiny teams and minimal funding. He now advises consumer mobile companies and runs advisory sessions through Intro, working with roughly 35 companies simultaneously.

Key ideas

Find latent demand in distorted behaviour. The insight for tbh was the number-one US App Store app being entirely in Arabic — proof that people wanted something so badly they were using a misfit tool. When users go through a distortive process to obtain a value, that value is the product.
Teens invite people at the highest rate; every year of age above 13 reduces it by 20%. If you build a social product for adults, expect to acquire every user with paid ads. The economics only work if the network effect is stronger than the CAC curve.
Test one condition at a time. Seed a school or community to eliminate the “not enough friends” confound, then measure whether the core loop works. Move to the next conditional only when the current one is proven.
Products live and die in the pixels. For zero-to-one consumer products, the PM should own the hierarchy, the flows, and the design. At large companies, PMs are typically separated from these decisions; this is the structural reason large companies struggle with new consumer products.
Three reasons people download apps: to make or save money (WhatsApp, PayPal), to find a mate (Tinder, Snapchat), or to escape reality (Netflix, Fortnite). Being honest about which motivation a product serves is the prerequisite for honest product development.

Episode content

The founding history: Midnight Labs

Nikita started building consumer apps seriously after Politify — a web app that calculated the financial impact of presidential policy proposals and went viral in 2012. He co-founded a studio called Midnight Labs (later called Outline) and over four to five years built approximately 15 apps, covering nearly every consumer category: maps, chat, events, to-do lists. Most were failures.

The studio model was the education. Each failure taught something specific about mobile mechanics, audience behaviour, or viral distribution. By the time of tbh, the team could build an app in two weeks (versus a year for the first) and had a repeatable process for testing whether an app had a shot.

Why teens

The decision to focus on teens was empirical, not ideological. After years of building for older audiences, the pattern was clear: organic growth only happened in high-density social environments (schools) where people saw each other daily and invited each other frequently.

Key observations:

Adults almost never invite friends to new apps. After age 22, invitations drop close to zero and paid acquisition becomes the only lever.
Teens see each other every day. This is the critical variable. A social app that requires someone to remember to invite a friend when they next see them fails; an app that is relevant to people you are physically with right now succeeds.
An academic study tracked the number of distinct people users texted from age 13 to end of life. The number peaks around 21, then falls sharply as people exit college, marry, and reduce their social surface area. The implication: target the ascending part of this curve.

tbh: the origin and the spread

The insight for tbh was a combination of observations:

A teen informant described a game called “TBH” happening on Snapchat: posting emojis and having followers reply with what they meant for you. Teens were improvising a positivity mechanism.
The number-one app in the US App Store was Sarahah, an anonymous messaging app from Saudi Arabia — the entire app was in Arabic. Millions of Americans were using an Arabic app because it was the closest available tool for receiving anonymous messages.
Most anonymous apps produced negative messages and bullying. tbh’s design insight: what if the only possible messages were positive, authored through polls rather than typed?

The combination — teens want to know good things about themselves, the closest existing tools are misfit or harmful — defined the product.

Spreading within a school: the team’s seeding tactic was to seed a single school (Georgia, chosen for the earliest school year start date) by targeting it on Instagram (high schoolers put their school in their bio, making them identifiable) and following them to get followbacks, then messaging them about the app. This was not how tbh grew — it was how they tested whether tbh worked. Once 40% of the first school had the app in 24 hours, Nikita knew the product had merit.

Growing between schools: organic. A student at the seeded school told a friend at another school; the friend downloaded it and found friends on it; the process repeated. Nikita geo-fenced the app to control the server load during this spread.

At peak: 360,000 installs per day. The servers cost $120,000/month; the bank account held $150,000. Nine weeks after launch, Nikita ran a competitive auction between acquirers and sold to Facebook.

Working at Facebook

Nikita was placed as a product manager on the youth team. His assessment: at a large tech company, PMs are separated from the actual product decisions. They are not in the pixels. Design is a vertical org. Data science is a vertical org. The PM coordinates and approves, but does not design the hierarchy or the flows.

His view: this structure makes sense for large, well-understood products where coordination is the scarce resource. For zero-to-one consumer products, it is fatal. You cannot verbally pitch a teen social mechanic in a VP review meeting and expect it to survive. The idea has to be in the pixels for anyone to evaluate it.

One structural observation: large companies take 12 to 24 months to respond to competitive threats in the consumer market. A PM posts about a new App Store chart-topper; the market research team does a study; a framing deck goes to VP review; development starts. By the time the response ships, the window has closed. This is not incompetence — it is the incentive structure of large organisations, where defending a failed bet is harder if the bet was not based on hard evidence.

Gas: rebuilding the viral mechanic

Gas was essentially tbh rebuilt for 2022 conditions. The core mechanic — anonymous positive polling — was the same. Almost everything else had changed.

The biggest structural difference: Twilio-based text invitations (which drove tbh’s spread) were no longer legally permissible for server-side sending. Gas had to rebuild the invite mechanism from scratch, using device-to-device invitations (slower, lower conversion). This alone required nine launches and multiple app rebrandings before the viral loop worked reliably.

The human trafficking hoax was a separate crisis that overlapped with the viral scaling. Entirely fabricated, it spread faster than the app at its peak: 3% of users deleting their accounts per day. Nikita fought it on every vector simultaneously — working directly with TikTok leadership to remove videos, insisting The Washington Post headline read “Gas is NOT for human trafficking,” calling police chiefs and school superintendents to get public retractions, and launching a TikTok video from his girlfriend that appeared in the deletion flow. The deletion rate fell to 0.1%.

Product principles Nikita extracted

Latent demand signal: look for users who are trying to obtain a value through a distortive process. The worse the workaround, the stronger the latent demand.

Eliminate confounds when testing: before concluding a product does not work, ensure the failure could not be explained by insufficient initial density. Seed aggressively, then measure.

Conditional product development: frame the problem as “if X is true, then Y must also be true for this to work.” Identify the four or five conditions that must all be true, prioritise them by uncertainty and consequence, and validate them in order. Do not build the whole product before validating the first condition.

24/7 live chat: put a live human in customer support during tests. It eliminates the “did they get help?” confound and, more valuably, generates the best feedback you will ever receive — users telling you in real time what is broken.

Inverted time-to-value: get users to the aha moment in seconds, not onboarding steps. For Dupe (a product he advised), the aha moment was typing “dupe.com/” before any product URL to find a cheaper alternative. The entire product experience was the aha moment itself.

Marketing and product are the same thing: the funnel from ad to in-app experience must be a single coherent promise. An ad that reaches a community and an onboarding flow that does not immediately connect that user to their community creates churn before activation.