Intrapreneurship: Kickbox, 15% Time, and the Myth of the Corporate Rebel

In 1982, Hewlett-Packard gave Chuck House a literal medal — the “Medal of Defiance” — for continuing work on a display monitor after David Packard had told him to kill it. The monitor became a hit, the medal became a slide in every innovation keynote, and the moral everyone took away was: real innovators defy management. Find your rebels, protect them, get out of their way.

I want to talk you out of that moral, because I’ve watched it do damage. The rebel story is survivorship bias in its purest corporate form. For every Chuck House there are dozens of people who defied a kill decision, were wrong, and were managed out — and they don’t get medals or keynote slots, so the sample we learn from is exactly the sample that flatters defiance. Worse, the rebel framing lets the organization off the hook: if innovation comes from special personalities, the company’s job is just to hire mavericks and tolerate them, and no process has to change. Notice that even the House story, read closely, is a system story — HP’s culture made the defiance survivable, and Packard handed out the medal himself. The defiance was absorbed, not punished. That’s a property of the institution, not the individual.

This series has covered why good companies kill good ideas, the structures that let exploration coexist with execution, and how to run a lab as a school. But structures and labs are containers. Someone still has to do the exploring, and most of them work nowhere near a lab. The question this post answers is: what does a company actually build so that ordinary employees — not anointed rebels — can take an idea from hunch to evidence? The honest answer, across forty years of programs, is three mechanisms: slack-time policies, seed-kit programs, and internal venture paths. Each one works when it makes permission cheap and evidence mandatory, and decays when it makes either expensive.

The word was coined to sell a system, not a personality

Start with the vocabulary, because it’s been bent. “Intrapreneur” entered the language through Gifford Pinchot III — a 1978 paper with Elizabeth Pinchot, then the 1985 book Intrapreneuring — and the popular memory of the book is the rebel memory: dreamers who do, forgiveness over permission, come to work every day willing to be fired.

But Pinchot’s actual argument was institutional. Most of Intrapreneuring is addressed not to the intrapreneur but to the corporation, and its load-bearing ideas are system ideas: intrapreneurs must be volunteers, not appointees; the person who has the idea should be the person who builds it, because handing a hunch to a “more qualified” team strips out the tacit knowledge that made it worth pursuing; and the company owes its intrapreneurs discretionary resources and staged commitment, not a one-shot pitch to a committee. He was describing, in 1985, roughly what Kickbox would ship in a box in 2013. The rebel quotes survived; the operating manual got forgotten. That’s a pattern this post will repeat several times.

15% time and the twelve-year Post-it

3M’s 15% time is the oldest slack-time policy still running, and the Post-it note is its sacred story — usually told as a eureka: scientist invents weird glue, colleague sticks it on a hymnal bookmark, office supply immortality.

The real timeline is the interesting part. Spencer Silver created his low-tack, reusable adhesive in 1968 — while looking for a stronger adhesive — and then spent years giving internal seminars trying to find a problem for his solution, earning the half-affectionate title “Mr. Persistent.” Art Fry, who heard one of those seminars, had the bookmark insight in 1974. The first product test-marketed in 1977 as “Press ‘n Peel” and flopped; only after a saturation sampling campaign in Boise did the relaunched Post-it note go national in 1980. From adhesive to product: about twelve years, spanning a failed launch.

So the system’s real feature was never the 15% of a week. It was tolerance for long gestation — an institution in which a researcher can keep an unjustifiable thread alive for years, keep talking about it internally, and find his collaborator through a seminar rather than a staffing decision. That tolerance was policy, not accident: William McKnight, 3M’s long-time chief, had written it down back in 1948 — hire good people, leave them alone, and accept that “management that is destructively critical when mistakes are made kills initiative.” The 15% number is the merchandising. The gestation tolerance is the mechanism, and it’s the part copycats never copy, because it doesn’t fit inside a fiscal year. Hold that thought; it’s where this series ends.

20% time, and the honest coda nobody puts on the slide

Google’s 20% time is the modern version of the story, blessed in the 2004 founders’ IPO letter: engineers encouraged to spend a fifth of their time on projects they think will benefit Google. The trophy case is genuine — Gmail grew out of Paul Buchheit’s side explorations, Google News out of Krishna Bharat’s, and the origins of AdSense are tangled up in the same slack. For a while, 20% time was the single most cited piece of evidence that giving smart people slack produces compounding returns.

Here’s the coda. Marissa Mayer said the quiet part while still at Google: it was really “120% time” — the side project came out of your nights and weekends, on top of a full load. By 2013, reporting (Christopher Mims’s much-argued-over piece, and plenty of internal accounts since) described the policy as effectively dead: projects needed manager approval, and time spent on them counted against you in a performance system that measured you on your team’s committed goals. Google never formally cancelled 20% time. It didn’t have to. It just let two other systems — resource approval and performance review — quietly reprice it until the cost of using the permission exceeded what most careers could afford.

That’s the diagnostic lesson, and it generalizes: a slack-time policy without a funding path and without protection from manager incentives is a press release, not a system. The 20% was real permission with no downstream anything — no staged money when an experiment showed signal, no way for your manager to get credit for your exploration instead of paying for it. 3M’s version survived because the surrounding institution metabolized long-gestation work. Google’s decayed because the surrounding institution measured quarterly and promoted on shipped commitments. Same policy on paper; opposite fates; the difference was everything around the policy.

Kickbox: permission as a physical object

Which brings us to the best-designed intrapreneurship mechanism I know of, because it was engineered directly at the failure point. In 2013, Adobe’s Mark Randall started handing any employee who asked a red cardboard box: a six-level guide for taking an idea from instinct to evidence, some caffeine and sugar for morale, and — the part that matters — a prepaid credit card with $1,000 on it. No application. No committee. No manager sign-off. Roughly a thousand boxes went out, around a tenth of the company at the time; at least 23 projects earned follow-on investment (the “blue box” stage, where an executive sponsor and real funding enter); and Adobe open-sourced the whole kit so anyone can run it.

The first post in this series located the kill point for internal ideas at the middle layer — the manager whose rational incentives make “no” the safe answer to any proposal that’s small today. Kickbox is a precision strike on exactly that veto. The $1,000 card means the first experiments need nobody’s yes: the permission conversation that used to be a social negotiation with your boss becomes a physical object you pick up. And $1,000 is a deliberately absurd sum — far too little to build anything, exactly enough to test something: run ads against a landing page, interview customers, buy the components for an ugly prototype. The constraint is the curriculum. The box doesn’t fund your idea; it funds the evidence about your idea.

Which is why I keep insisting Kickbox is a pedagogy and a funnel wearing a gimmick’s costume. A thousand employees worked through a self-paced course in validating ideas — that’s the school thesis from the labs post, distributed in cardboard. And the red-box-to-blue-box progression is staged funding in miniature: cheap, universal permission at the bottom; scarce, evidence-gated commitment above. Most ideas die at level three or four, killed by their own founders’ data, which is the funnel doing its job. The company’s real cost per killed idea is a thousand dollars and some candy. Compare that to what your last zombie project cost.

Lean startup goes inside: Intuit, and the GE warning label

The methodology underneath all of this — hypothesis, cheapest possible test, evidence before scale — is the lean startup toolkit, and the 2010s produced two giant enterprise deployments worth reading together.

Intuit is the durable one. Scott Cook and Brad Smith spent years wiring “Design for Delight” and rapid experimentation into the operating culture: unstructured time for employees, a network of trained innovation catalysts coaching teams through experiments, and leadership rhetoric that Cook summarized as replacing decisions by persuasion with decisions by experiment — the boss’s opinion loses to the customer’s behavior. The results that came out of that machinery weren’t lab products; they came from working teams. Fasal, a mobile price-information service for Indian farmers, grew out of employee experimentation in a market Intuit had no business plan for. SnapTax came from a small team’s bet that a tax return could be filed from a phone photo. Neither needed a rebel; both needed cheap experiments and a leadership that had pre-committed to respecting the results.

GE FastWorks is the cautionary tale, and it deserves a careful reading rather than a smirk. Starting in 2013, GE hired Eric Ries to help retrain the company — tens of thousands of employees through FastWorks programs, the largest lean-startup deployment ever attempted, with genuine wins like the Series X engine program cutting projected development time and cost. Then GE collapsed — and the collapse became, in conference-hallway shorthand, evidence that lean startup doesn’t work in the enterprise.

Be precise about what GE’s fate does and doesn’t prove. GE was brought down by GE Capital’s buried risks, disastrously timed bets in power, dividend and accounting pressures years in the making — causes that predate FastWorks and run entirely above the altitude where any product methodology operates. FastWorks didn’t fail GE; it was irrelevant to what killed GE, which is a different indictment. What the story does prove is about limits: teaching thirty thousand people to run experiments changes nothing if the capital-allocation system above them still funds annual plans and punishes the variance that experiments exist to surface. A methodology adopted underneath an unchanged funding and measurement system is a coat of paint on a load-bearing wall. Intuit changed the wall. GE painted it at unprecedented scale.

A venture path is not a suggestion box

The last mechanism is the one that separates companies that harvest intrapreneurship from companies that merely host it. Most organizations have a suggestion box in some costume — an idea portal, an annual hackathon, an innovation jam. Ideas go in; badges, gift cards, and silence come out. A venture path is a different animal, and you can tell them apart by three tests.

Staged funding that’s real. Not a one-time prize, but tranches released against evidence, discovery-driven-planning style — each gate re-decides based on what the last tranche’s experiments showed. Kickbox’s red-to-blue progression is the entry-level version; a real venture board reviewing a portfolio quarterly is the grown-up one.

The founder stays on the venture. This is Pinchot’s 1985 rule, still routinely violated: the moment an idea shows promise, the org “staffs it properly” and reassigns the originator, destroying the tacit knowledge and the obsession in one move. A venture path lets people follow their idea out of their day job — which also quietly solves the 120%-time problem, because exploration stops being an unfunded second shift.

A declared endgame. Absorption into a business unit, independence as a spin-out, or an honest kill — decided by evidence at a gate, not by orphaning. This is where the ambidexterity sponsor from earlier in the series earns their title: someone senior enough to own both the core and the venture has to force the absorb-or-spin decision, because no middle layer will volunteer to inherit something that dilutes next quarter’s numbers.

Run the three tests on your own company’s program. If ideas enter but no money is staged, no founder is released, and nothing has ever been absorbed or spun out, you have a suggestion box with a marketing budget — a morale program pretending to be a strategy.

The part none of this survives

Here’s the uncomfortable summary of forty years of evidence. The mechanisms are known and even open-sourced: slack time works when the institution tolerates long gestation; seed kits work when they delete the middle-manager veto; venture paths work when funding is staged, founders stay, and a sponsor forces the endgame. None of this requires rebels. All of it requires ordinary people plus a system in which trying is cheap and evidence, not seniority, decides what continues.

And every one of these mechanisms dies the same death: annual budgeting. 20% time died when performance review repriced it. FastWorks bounced off a capital-allocation system it was never allowed to touch. Even 3M’s gestation tolerance is, at bottom, a funding posture — a willingness to carry unjustifiable work across fiscal years. You can hand out red boxes forever, but the moment a validated idea needs its second and third tranche, it collides with a planning process built to fund certainty and measure everything on this year’s P&L. That collision — how to fund and measure work that doesn’t fit a P&L yet — is the last and hardest piece of innovating from within, and it’s where the final post in this series goes next.

Further reading

• Gifford Pinchot III, Intrapreneuring (1985) — read past the rebel quotes for the institutional operating manual everyone forgot.
• Adobe’s open-sourced Kickbox kit — the complete six-level curriculum and the reasoning behind the $1,000 card, free to run in your own company.
• Eric Ries, The Startup Way (2017) — the FastWorks and enterprise-deployment account from the inside; read it against GE’s subsequent history for the full lesson about methods versus funding systems.