Proven Better New

A framework for turning an instinct into a product

Mark Pincus, the founder of Zynga, describes a three-part routine he uses to move from a raw instinct to a product without betting everything on one untested guess. He frames the posture first: approach the idea "like a scientist with a white lab coat," dispassionate about your specific ideas even while remaining passionate about the underlying instinct.¹ The three sequential questions are Proven, Better, and New.

Proven. Pincus starts by finding whoever has done the closest thing to the intended product and deconstructing every element that drove its success. He applies two filters. The comparison has to be for the exact same audience, because proof against a different customer is weak data. And failed products count as evidence too, since, in his words, they often contain the right instinct hidden in the wrong execution.

Better. Before inventing anything, he looks for the improvement that ten out of ten of the existing audience would immediately recognize as better. The emphasis is that this is what they think is better, not what the founder thinks. Pincus notes these wins are usually small and mundane, such as free, half price, no download required, or a better soundtrack, and that the bar is a "fuck yeah," not a polite "that's interesting."

New. Only now does the novel differentiator get articulated, and Pincus attaches a counterintuitive instruction to it: expect it to fail. "The New will probably fail. It'll be a reason to try it and it will probably fail." Its job is to be a reason to try the product and to be an isolated variable, so that if the product does not work, the team can tell whether the novel bet was wrong or whether a foundational problem should have been caught earlier.

What the foundation buys

In Pincus's account, doing Proven and Better well purchases two things. First, the product will not fail for the wrong reason, because a solid foundation makes any failure informative and isolates the New as the variable that moved. Second, it protects the instinct: if the New fails but the foundation holds, the instinct is likely right and only the implementation of the novel element needs to change. This is why he reframes "fail fast" into something more precise, failing informatively and cheaply rather than iterating at random.

He also names a common failure mode he calls the buried gem: a product that flops commercially while containing a genuine innovation, because the team did New first without Proven or Better, so no one could find the good part. The framework implies giving yourself several parallel bets rather than marrying the first hypothesis: "don't just stick with the one shot where you're like, okay, is it working? I don't know."

Applied examples

Pincus points to his own Zynga practice: "I had a whiteboard, I just wrote the name of every game mechanic I'd ever heard of. I would try them and cross them out as we did them." That whiteboard is the systematic version of testing isolated New variables on top of a Proven and Better foundation of social-network game mechanics. He describes the same routine applied secondhand to Nikita Beer's app TBH, which copied a proven concept, an anonymous-honesty-box mechanic, from a niche Arabic-language app, made it front and center instead of buried in a narrow niche, then layered teen social affirmation mechanics as the New.

Where it strains

Pincus's own qualifications are worth carrying forward. Proven and Better are backward-looking by construction, so in a rapidly shifting market the historical comparison may not be informative. The framework also assumes the exact audience is known in advance, which can exclude products that only discovered an unexpected audience mid-iteration. And running several parallel New tests demands capital and bandwidth that early teams rarely have. Pincus treats the discipline as a way to keep conviction at the instinct layer while staying ruthlessly experimental at the idea layer.