How to kill bad ideas and stay true to winning instincts

Executive overview

Most entrepreneurs cling to bad ideas too long, burning time and capital on execution that was never going to work. Mark Pincus's career teaches a sharper discipline: separate the instinct (usually right) from the idea (often wrong) and kill the idea fast.

Zynga became a hit factory not by having better ideas than competitors, but by testing hundreds of ideas daily and terminating failures before they drained resources. The framework is guerrilla testing, data standardisation, and relentless iteration — all in service of the customer experience.

The instinct behind an idea is often correct even when the idea itself is wrong — kill the idea, not the instinct.

The cost of not killing ideas: lessons from Tribe

• Tribe (2003) was a social network focused on urban subcultures and interest groups.
• Three instincts embedded in Tribe were all individually correct: real-name social networking (Facebook), reputation-driven sub-communities (Reddit), professional networking (LinkedIn).
• The single bad idea was combining all three into one product open to anyone, including strangers.
• Tribe became associated with niche subcultures and failed to attract mainstream users.
• Pincus's own girlfriend refused to use it after receiving unsolicited messages — a focus group of one that turned out to be more accurate than his data.
• His diagnosis: "As entrepreneurs, part of our journey is learning how to separate our winning instincts from our losing ideas."
• Rule of thumb: a good entrepreneur's instincts are right ~95% of the time; their individual ideas are right ~25% of the time.

Zynga's idea-killing machine

• The core insight from Tribe: stop being wedded to any single idea; try anything, kill anything, kill it quickly.
• Zynga's founding thesis: massive latent demand for games existed, but friction (downloads, hardware requirements, scheduling) blocked it. Facebook's third-party app platform removed that friction.
• Traditional game studios spent 2 years and $15–20M before knowing if anyone wanted a game. Zynga could run a test in four weeks.
• For every hit (Farmville, Zynga Poker, Words with Friends) there were far more ideas that never shipped.

Guerrilla testing

• Zynga tested ideas before building them: a five-word prompt inside an existing game described a hypothetical new feature.
• If users clicked, the response was "Coming soon — want to be notified?" — no feature was built yet.
• Threshold for proceeding: if roughly half the audience clicked, the idea was worth pursuing.
• Tests ran across hundreds of ideas every day, covering small changes (button colour) and large ones (selling virtual in-game items).
• Ideas that cleared the threshold became bold beats — features designed to generate visible excitement among players.

Bold beats in practice: the moving cow

• An engineer made a single low-effort change: animals in Farmville could now move.
• The change lit up Twitter and the blogosphere without any marketing.
• It cost a few hours of engineering time and sparked months of follow-on features: animal breeding, horse racing, hatching eggs.
• Metrics and social signals spiked the moment the feature went live — immediate, unambiguous signal.
• "It was like Christmas Day because when we turned it on, your metrics would light up."

Data standardisation as a scaling tool

• As product teams multiplied, each measured differently — making cross-team learning impossible.
• Pincus mandated a single measurement template across all teams: same metrics, same test formats.
• Resistance was high; he overrode it — "you can fight like hell for what you think is right, but then we're all going to commit to it."
• Result: a shared cache of insights that every team could read and act on.
• Zynga hired data engineers and analysts at an accelerating pace when competitors were hiring game developers — widely mocked at the time.
• Over-investing in data gave Zynga the ability to act on findings faster than any competitor could comprehend.

Customer experience above all else

• As Zynga scaled, Pincus's core warning was: don't let the mechanics of scaling distract from delighting the customer.
• Users don't care how a great experience was delivered — only that it was.
• Design and product decisions matter at the pixel level, especially on mobile where every pixel and every second counts.
• Micro-manage the things that drive customer excitement; standardise everything else.
• Instil that obsession across the entire company so it survives scale.