Ch8 02: Stage Gates for Startups: How to Kill Bad Ideas Before They Kill You#

I’ve watched smart, experienced leaders pour millions into projects that everyone in the room privately knew were failing. Not because they lacked information. Not because they couldn’t do math. But because stopping felt like admitting defeat, and admitting defeat felt worse than writing another check.

This is the sunk cost trap — one of the best-documented cognitive biases in psychology, and one of the most destructive forces in business. The more you’ve invested in something, the harder it is to walk away, even when every rational signal screams that you should.

In innovation, this trap is lethal. Because innovation is inherently uncertain, every project will hit a stretch where the evidence is murky. Is this a rough patch or a fatal flaw? Push through or pull the plug? Without a structured decision framework, the default is almost always to keep going. “We’ve come this far.” “Just a bit more time.” “The next milestone will prove it.”

Those sentences have consumed more capital than any competitor ever has.

At DVx — the venture studio I run — we deal with this by taking human judgment out of the continue/kill decision. Not completely, but enough to counterbalance the bias.

The tool is called the stage-gate system. Here’s how it works: before a project launches, we define a series of gates — specific milestones with specific, measurable criteria. Each gate is a decision point: does the project move to the next stage, or does it stop?

The criteria are locked in before work begins — not later, when emotions are running and sunk costs are stacking up. They’re as concrete and quantifiable as we can make them. “Has the team validated willingness to pay with at least fifty customers?” “Is the unit economic model positive at current scale?” “Has the technology been demonstrated in a production environment?”

At each gate, the project is measured against the pre-set bar. Pass, and it advances. Miss, and it stops. Not “pauses.” Not “pivots into something vaguely adjacent.” Stops. Resources get freed and pointed at the next opportunity.

DVx’s gate system has five stages:

Gate 0: Problem validation. Does a real problem exist, and is it big enough to matter? The test isn’t “do we think it’s a problem” but “have we found people actively trying to solve this problem and failing?” Real problem, project advances. Solution in search of a problem, it dies here.

Gate 1: Willingness to pay. Will people open their wallets? Not “would they theoretically pay” — have they actually put money down, even a small amount, on a prototype or pre-order? This gate kills the most common innovation failure: building something people say they want but won’t pay for.

Gate 2: Technical feasibility. Can the solution be built at a cost that supports a sustainable business? The bar isn’t “can we build it” (almost anything can be built) but “can we build it at a cost that leaves room for profit?”

Gate 3: Market validation. Can we acquire customers at a cost that scales? Customer acquisition cost, retention rate, and lifetime value must clear pre-defined thresholds. If the math works small, advance. If acquiring customers takes heroic effort, the economics won’t magically improve at larger scale.

Gate 4: Scale readiness. Can the org, supply chain, and tech handle ten times current volume? This gate ensures that scaling doesn’t shatter what worked at small scale.

The most counterintuitive part of this system is that it boosts innovation rather than smothering it.

The standard pushback: “Gates kill creativity. If Edison had a stage-gate system, he’d have quit after the first hundred failed filaments.” Sounds sharp, but it misreads what the gates measure.

Gates don’t measure effort or creativity. They measure evidence. “Did the customer pay?” isn’t a creativity test. It’s a reality test. And the projects that clear reality tests are the ones that deserve more creative energy — not less.

Here’s what actually murders creativity: resource starvation caused by zombie projects. When an organization’s innovation budget is locked up in three initiatives that should’ve been killed six months ago, there’s nothing left for the next great idea. The gate system doesn’t shrink the total volume of innovation. It increases innovation per dollar by routing resources to the projects with the strongest evidence, not to the projects with the loudest advocates.

Think of it as an investment portfolio. No VC expects every bet to pay off. The goal isn’t a 100% hit rate — it’s a portfolio return that justifies the total outlay. Some projects fail. Expected. What’s unacceptable is continuing to fund failures at the expense of potential winners.

There’s one more benefit I didn’t fully appreciate until I’d run the system for several years: it makes failure safe.

In organizations without gates, killing a project is a traumatic event. A senior leader has to make a painful judgment call. The project team feels attacked. The leader feels like an executioner. Everyone dodges the conversation as long as possible, which means projects limp on long past their sell-by date.

With gates, termination isn’t a verdict on people. It’s a system output. The criteria were set up front. The data either clears the bar or it doesn’t. “The project didn’t pass Gate 2” is a statement of fact, not an indictment. The team can redeploy to a new project without stigma, because the system treats failure as data rather than blame.

That psychological safety — knowing that project failure won’t morph into personal failure — actually encourages bolder swings. When people know failure will be caught early and handled cleanly, they’re more willing to chase ambitious, high-risk ideas. The gates protect the downside, which frees people to go after the upside.

Guidance#

If you’re running innovation projects — startup, venture studio, or R&D inside a big company — put a gate system in place:

Set gates before the project kicks off. Not during. Not after. Before. Each gate should carry two to three measurable criteria the project must hit to advance.
Make criteria specific. “Customer traction” isn’t a criterion. “Fifty paying customers with monthly retention above seventy percent” is. Vagueness is the escape hatch for emotional decision-making.
Evaluate on schedule. Each gate has a date. When it arrives, the project is evaluated. No extensions. No “just one more month.” Fixed evaluation dates are where the system’s power lives.
Kill cleanly. When a project misses a gate, stop it. Redirect the resources. Celebrate the learning. Move on. The most expensive innovation isn’t the one that fails — it’s the one that should’ve been stopped but wasn’t.
Track portfolio return. Don’t judge individual hit rates. Judge the return on your total innovation spend. A portfolio where three out of ten projects succeed big and seven are stopped early beats one where all ten stumble along indefinitely.

Persistence is a virtue — but only when aimed at the right target. The stage-gate system helps you tell the difference between persistence that leads to breakthroughs and persistence that leads to bankruptcy.