Kill Discipline

Ettore, my youngest son, has started building block towers lately. Not very tall, to be honest - but still… He stacks them with intense concentration, tongue out, trying to adjust the angle of each block until the tower reaches whatever height satisfies him. Then he knocks it over. No hesitation. No committee meeting. He watches the blocks scatter across the floor, picks one up, and starts something new.

He’s fourteen months old. He has better kill discipline than most Fortune 500 companies.

In the first article in this series, I pulled apart an argument from an article published at Tom’s Hardware Italia, where it was claimed AI had made ideas officially worthless and only execution matters now. I wrote about the line between generation and judgment is dissolving from both directions. AI is colonizing judgment from above while arming fakers from below. The signals we used to rely on are degrading.

But before asking what replaces those signals, there’s a prior question worth sitting with: who built a professional ecosystem on the premise that generating ideas was the hard part? Because that ecosystem is enormous. And the incentive structure that sustains it explains why innovation theater keeps running long after everyone in the audience has figured out the plot.

Everyone Knows

Steve Blank named it in Harvard Business Review back in 2019. Companies adopt hackathons, design thinking workshops, and innovation sprints without building the processes to convert any of it into actual products. Leadership hires management consultants who reorganize without enabling. Everyone looks busy. The word “innovation” shows up in slide decks. Nothing ships.

The data backs him up. Fast Company found very little evidence of hackathons leading directly to major market successes. Participants often lack the contextual knowledge to produce feasible solutions. The format privileges speed over depth and novelty over viability. Companies like GM and GE ran them anyway. The results “very rarely spark real, lasting innovation.”

But the most telling example comes from Duke Corporate Education. They documented a major consumer goods firm that invested in more than 150 teams across three months of incubation. At the end of the program, the teams returned to their day jobs. No funding, no follow-up. A hundred and fifty teams generating ideas that went exactly nowhere. At a different company, a leader told his team to bring him “ten times the current level of ideas,” as though the constraint on innovation were a shortage of Post-it notes.

And the underlying data is worse than the anecdotes suggest. Viima’s analysis put numbers on it. Brainstorming, it turns out, doesn’t generate better or more ideas than when people work alone. And organizations measure the wrong thing: number of ideas generated, workshops held, hackathons organized, participants counted. Input metrics where outcome metrics should be. They measure motion and call it progress.

None of this is news. If you’ve spent any time in mid-to-large organizations, you’ve lived some version of it. You’ve stood in a room full of Post-it notes and felt the gap between the energy in the workshop and the silence three months later when nothing happened.

To be fair, hackathons can work. Recent academic research shows that when companies design them with strategic follow-up processes, dedicated funding for winners, and post-event integration plans, the format produces real value. But that’s the point. The scaffolding is what turns activity into outcome. And the scaffolding is exactly what innovation theater skips.

The interesting question isn’t whether innovation theater exists. Everyone knows it does.

The interesting question is why it survives.

Follow the Money

The global management consulting industry brings in somewhere north of half a trillion dollars a year. McKinsey alone reported over $16 billion in 2024. Innovation consulting is one of the fastest-growing segments. That growth did not happen because ideation workshops produce results. It happened because ideation workshops are easy to sell.

Think about the product from the consulting firm’s perspective. An ideation workshop is a two-day engagement with clean deliverables. There are Post-it notes on a wall. There’s a photo with the CEO. There’s a deck with “top ideas” ranked by some criteria the consulting team designed. The client feels something happened. It’s repeatable and scalable. And let’s be honest - it can be a very fun and engaging activity.

Now try selling the opposite. Try walking into a boardroom and pitching: “We’re going to help you kill projects.” See how that meeting goes. Project termination requires months, not days. It means telling executives their initiative isn’t working. It means unwinding budgets, reassigning teams, and writing a narrative that doesn’t sound like failure.

The deliverable is a decision that most organizations are psychologically incapable of making on their own, because fifty years of behavioral research, starting with Barry Staw’s work on escalation of commitment in 1976, has documented exactly why. Sunk costs make people throw good money after bad. Face-saving makes them frame failure as “pivoting.” And the accountability structures inside most companies reward the person who keeps a project alive far more than the person who kills one.

The consulting industry didn’t create “innovation theater” out of malice. Nobody sat in a conference room and decided to sell clients something useless. The problem is that the end product itself happens now to be cheaper, more abundant. Porcu was right about that much: ideas were always the easy part. The consulting industry just built a business model around that. And AI just made it even cheaper.

PwC’s 2026 AI Business Predictions show the same pattern reproducing. Companies crowdsource AI initiatives from the bottom up, then try to assemble the results into something resembling a strategy. PwC’s assessment: “impressive adoption numbers” but projects that “may not match enterprise priorities, are rarely executed with precision, and almost never lead to transformation.” The format changed from brainstorming to prompt engineering. The incentive structure stayed the same. Sell the activity. Bill for the motion. Leave before anyone asks what the outcomes were. The fact that an AI can now generate a hundred product ideas in thirty seconds should have killed the ideation workshop market overnight. Instead, it just gave the workshops a new prop. And also more value to a well structured funnel to select the best ideas to bring to life.

The Skill Nobody Sells

The organizations that recognize the problem still tend to misdiagnose the solution. They create new roles instead of building new capabilities.

Ayelet Israeli and Eva Ascarza wrote in Harvard Business Review last November that most AI initiatives fail not because the models are weak but because organizations aren’t built to sustain them. “Technology enables progress,” they found, “but without aligned incentives, redesigned decision processes, and an AI-ready culture, even the most advanced pilots won’t become durable capabilities.” The teams resist. The behaviors don’t change. The pilot works in a sandbox and dies in production. Not because the technology failed but because nobody redesigned the organizational plumbing around it.

Thomas Davenport and Randy Bean reported the symptom from the other direction. Thirty-eight percent of companies have now appointed a chief AI officer or equivalent. But there’s little consensus on where that role sits: some report to technology, some to business, some to transformation, some to a chief data officer. The reporting structure is so fragmented that Davenport and Bean blame it directly for “the widespread problem of AI not delivering sufficient business value.” Creating a title isn’t building a capability. It’s another form of theater. New label, same plumbing.

The real gap isn’t that organizations lack ideas. Most of the times they don’t even lack selection processes. What they lack the most is the organizational muscle to stop. To terminate a project that isn’t working, redirect the resources, and treat the termination as information rather than failure.

Venture capital has this muscle. It’s built into the structure. Seventy-five percent of VC-backed companies never return cash to investors. VCs don’t treat that as a catastrophe but rather as the cost of running a portfolio. Staged financing means every funding round is a kill decision: continue or stop. The default is termination unless the data justifies more investment. Ten to fifteen percent of bets produce the returns that fund the whole portfolio.

Corporate innovation operates on the opposite default. The default is continuation. Projects survive because nobody wants to be the person who killed them. The political cost of stopping is higher than the financial cost of continuing. So things drift. Zombie projects sit on roadmaps consuming budget and headcount that could go somewhere useful. And the organization responds the only way it knows how: by generating more ideas. Because ideation is the one activity that doesn’t require anyone to say no.

I think this is where AI could play a different role than the one it’s currently being sold for. Not generating more ideas (that’s the part that’s already too cheap to matter) but pattern-matching against historical failure data to flag projects that look like past failures. The AI wouldn’t make the kill decision. It would give cover to the human who needs to make it but can’t do it alone. Not replacing judgment, which is a thread I followed in the first article, but supporting the specific kind of judgment that organizations punish rather than reward.

Nobody is selling that capability in a two-day workshop. Which is precisely why it might be the thing worth building.

What Ettore Knows

Ettore knocked over his tower again this morning. He looked at the pile of blocks on the floor, picked up two of them, and started building something different. No sunk cost analysis. No face-saving presentation to the living room about how the tower’s strategic direction had evolved. He was done with it, so he stopped.

I watched him and thought about those teams, those people at the companies I read about while doing the research for this article. Months of incubation. Zero follow-up. These companies didn’t lack ideas nor the people willing to generate them. What it lacked was someone willing to look at the program after month one and say: this isn’t going to produce anything, and we should stop now and spend the remaining budget on the three ideas that actually have a path forward.

The organizational equivalent of Ettore’s instinct is the most absent capability in corporate innovation. The willingness to stop. To say: this isn’t working, we’ve learned what we’re going to learn, and the resources go somewhere else now.

You will not find in the market that skill. It can’t be packaged into a two-day workshop. It requires redesigning incentive structures and accountability frameworks, changing what the organization treats as a successful outcome. That work is slow and political. It produces no Post-it notes and nice-looking decks for senior executives.

But the question behind this one is harder still. If the skills that matter now are judgment, coordination, and the willingness to stop, then who gets to develop them? Because the answer to that question is less democratic than the “just execute” crowd wants to believe. It’s the subject of the next article.