Proven, Better, New: Mark Pincus on the Rules of Product Innovation

What It Covers

Mark Pincus, founder of Zynga, shares the product development framework he calls "Proven Better New," built from failures at Tribe.net and successes with FarmVille. He covers how to identify genuine product-market signal, build rapid testing machines, maintain founder control, and avoid the MVP trap that kills most consumer products.

Key Questions Answered

• •Proven Better New Framework: Before innovating, become a PhD in what already works. "Proven" elements should be legally copied without modification — don't change what you don't understand. "Better" is typically mundane: free, no download, half the price. "New" is your single isolated innovation. Zynga Poker copied real poker tables exactly, removed the download requirement, then added one new element: real friend photos around the table.
• •True Signal vs. False Signal: Genuine product-market fit feels unmistakable — Pincus calls it "heat." When you have it, every metric confirms it without needing interpretation. When you don't, you find yourself assembling statistics trying to manufacture conviction. The practical test: if a consumer app earns a permanent spot on your phone's home screen, it has billion-dollar retention potential. Most digital products never reach that threshold.
• •Failure Machine Over MVP: Replace the minimum viable product with a minimum idea state. You lose roughly 50% of users at every click and 90%+ at any download prompt. Test click-through rates on raw concepts before building anything. Words with Friends was projected to drop from $120M to $79M in revenue; weekly click-testing hundreds of ideas around player achievements reversed that to $180M in revenue.
• •All New Fails Until It Doesn't: Treat every new product feature as a default failure. This isn't pessimism — it's a testing discipline. Run many variants of each new idea across many new ideas simultaneously, looking for signal in small atomic units. The team that built fast-play mode for Words with Friends spent hundreds of engineering days on a feature only 1% of players clicked. Click-testing first would have revealed that immediately.
• •Founder Mode as Control Architecture: Founders should accept lower valuations in exchange for full voting control. Pincus had a handshake deal to acquire Supercell for $400M when Clash of Clans was ranked 25th in the App Store. His board blocked it. Supercell generated $500M in profit the following year. Retaining voting control means you can override board consensus on bet-the-company decisions — without it, you're executing someone else's strategy.
• •Tech Assistant as Leadership Multiplier: Modeled on Andy Grove, Bill Gates, and Jeff Bezos, the tech assistant role is distinct from chief of staff. One high-potential employee — ideally a smart misfit who doesn't fit conventional roles — shadows every meeting, absorbs decision-making patterns, and works on research projects. Bezos used this pipeline to develop his entire C-suite, including Andy Jassy. It scales leadership transmission without adding management overhead.