Roger Chen on testing a social app with ads before building

Rapid validation through ads and tappable prototypes

The central thesis of Roger Chen’s approach to building viral consumer apps is that the most dangerous period for a builder is the "dreadful" gap between having an idea and launching it. During this window—which can last weeks or months—founders often work in isolation, building features based on assumptions rather than evidence, only to realize upon launch that they have taken the wrong path.

To solve this, Chen advocates for a two-pronged validation strategy: using paid ads to simulate market interest and using high-fidelity, "tappable" mockups to solve UX issues before a single line of production code is written.

This is not about scaling; it is about feedback. By running small-scale ad campaigns on platforms like TikTok, a founder can immediately see which product concepts, user interfaces, or "wedges" (specific use cases) resonate with real people. Simultaneously, by moving beyond static design tools like Figma and into high-fidelity prototyping tools like Proto.io, builders can experience the "feel" of an app. This reveals whether a user flow is too long or an interaction feels clunky, allowing them to fix these issues in a "fake" environment. The goal is to bridge the gap between a conceptual idea and a polished product by failing fast and cheap in the prototyping phase rather than the engineering phase.

Validation: Using Ads as a Feedback Loop

For many builders, paid advertising is viewed strictly as a scaling mechanism—a way to pour fuel on a fire that is already burning. Chen argues that this is a fundamental misunderstanding of the tool's potential in the early stages of a product's life. In the beginning, ads are actually the fastest way to gather high-quality feedback.

When you are in the "pivot hell" phase of entrepreneurship, you might spend months iterating on a feature only to find it flops upon release. Chen suggests flipping the traditional model: instead of building the full product and then trying to find users, use small ad spends to test the concept first.

Roger Chen:

"In the beginning, like ads are really the fastest way to get feedback... we gain two things. Number one, we gain at least one format that could work. And two, we see if people like our product."

By running these tests, a founder can identify a "winning format" for their marketing and a "winning wedge" for their product. For example, during the development of the AI companion app Bro, the team used ads to test different user interfaces and concepts—such as a lock-screen presence versus a floating window. This allowed them to validate which specific interaction model people actually wanted before committing expensive engineering resources to build the wrong thing.

However, there is a clear boundary to this strategy. Chen notes that as an app scales, the cost of acquisition naturally rises, making ads an expensive way to grow a massive user base. They are a tool for discovery and validation, not a permanent solution for mass-market scaling.

The Prototype Gap: Proto.io vs. Figma

A major source of waste in app development is the "prototype gap"—the distance between a static design and a functional user experience. Designers use Figma to create visual layouts and design systems. While these look professional, Figma is largely static; you can see what a screen looks like, but you cannot truly experience how a user moves through the interface.

To close this gap, Chen’s team uses Proto.io to create "tappable" prototypes. These are high-fidelity mockups that act like a real app—you can tap through flows, trigger transitions, and simulate the actual experience of using the interface.

While Figma is excellent for visual layout, it often fails to capture the nuance of temporal interactions—how a menu slides, how a button responds to a press, or how a flow feels when moving between screens. A Proto.io mockup allows a founder to treat the interface as a living object. Even if the "app" is just a collection of programmed screens (similar to an advanced PowerPoint), the ability to physically interact with the flow is what reveals whether a user journey is actually functional.

This approach offers two primary benefits:

1. Reducing Unnecessary Engineering: By catching UX failures in a mockup, the team avoids writing code for flows that are ultimately too long, too confusing, or simply "unfun."
2. Personal Intuition: Chen describes a practice of carrying a "fake" version of the app (even a physical brick shaped like the phone) to mentally and physically inhabit the design. This focus on how the device feels in the hand ensures the product is optimized for the user's actual behavior, not just for aesthetic appeal in a design tool.

Social Growth: Relationship and Temporal Density

Building a successful social app requires more than just a good feature; it requires a specific type of mathematical density. Chen highlights two critical metrics discovered during the growth of the app Lobby: Relationship Density and Temporal Density.

Relationship Density

Social apps thrive when users are connected to others immediately. Chen cites a notable gap in retention rates based on how many friends a user has on their first day. In a specific market test in Israel, Lobby saw a large difference in success compared to other regions.

This data proves that "seeding" a user base with enough connections to create a network effect is a primary driver of long-term retention. Once that density is hit, the app becomes a self-sustaining ecosystem.

Temporal Density

Even with high relationship density, a real-time social app can fail if users are never online at the same time. Chen calls this "temporal density." If two friends are part of the same network but are never active simultaneously, the app feels dead.

To combat this, the team implemented "low-pressure" interaction mechanisms designed to force these moments of synchronicity without making them feel like a chore.

• The One-Minute Call: Instead of a high-stakes video call that feels like a "conference room" meeting, Lobby introduced a one-minute "roll call." This is a low-pressure, timed interaction that feels more like a game. It solves the "hostage" problem—the feeling that once you join a video chat, you are stuck in it for an hour.
• The Group Picture: To drive viral growth, they turned video hangouts into sharable assets. By prompting a "group boomerang" or photo during a session, they created a mechanism for "FOMO" (fear of missing out). Friends who were not in the hangout would see the shared media and feel a desire to join the next one, naturally increasing both relationship and temporal density.

Marketing and AI: Hook, Demo, and Context

When moving from product design to mass marketing—specifically on platforms like TikTok—the strategy shifts from "building" to "storytelling." Chen identifies a highly successful, albeit unpredictable, pattern: the Hook and Demo.

The TikTok Framework

Success on TikTok is "top-heavy," meaning it relies on high-view outlier hits rather than a steady stream of average videos. To land these hits, content must follow a specific structure:

1. The Hook (A Human Scenario): You cannot just show an app interface; people will swipe away. You need a real person presenting a relatable, human-centric scenario (e.g., "I accidentally sent an 'I love you' text to my crush").
2. The Demo (The Product Solution): Once the user is hooked by the scenario, you show the app as the tool that resolves or complicates that scenario.

From Content to Context

As Chen moves into the AI space with Bro, his philosophy shifts from using AI to generate content to using AI to generate context.

Most AI companions currently act as "black boxes"—users go into the app to talk to the AI, which often feels lonely or dystopian because it displaces human interaction. Chen’s goal is to use AI to augment social relationships.

Instead of an AI that writes a poem for you (content), he envisions an AI "buddy" that acts as a "yappy friend" in a group chat. This AI doesn't replace the humans; it acts as a "hype man" or a conversation starter. It inserts funny observations, provides "context," and keeps the topic moving when a human conversation starts to lag. This mimics the role of a "third person" in a social circle who keeps the energy high, effectively offloading the "thankless job" of carrying a conversation to an AI agent.

For consumer app builders

For builders working on projects like Twelve Letters or Day One, these insights suggest a shift in both the development and marketing workflows.

Prototyping Workflow Do not rely solely on Figma for your core user journeys. If your app relies on a specific emotional rhythm—like the "heart-to-heart" guidance in Twelve Letters—build a high-fidelity, tappable prototype in Proto.io. Experience the flow of the AI's prompts and the user's responses as if it were real. If the "vibe" feels off during a mockup, it will definitely feel off in the final product.

AI Interaction Design When designing AI-guided experiences, avoid the "black box" trap. Rather than making the AI a solo entity the user interacts with, design it to facilitate connection. For an app focused on legacy, the AI should act as a bridge—not just generating a letter, but creating the context that helps a user reflect and eventually share that narrative with their family.

Go-To-Market Testing Before committing to a large-scale influencer campaign, run "micro-experiments" with small ad spends. Test different "wedges." For Day One, is the hook "Learn AI in 7 days," or is it "The beautiful way to master the future"? Use the ads to let the market tell you which story stops the scroll.