Abstract

Innovation rarely fails for lack of ideas. It fails because the environment is hostile to ideas while they are still fragile. When constraints, skepticism, or institutional inertia arrive too early, two things happen at once: promising ideas get killed before they can prove themselves, and people stop offering ideas at all. They learn to wait to be told. Over time that is one of the quiet reasons good people leave. This paper examines how six companies solved that problem, each in its own way: Google, 3M, Amazon, Pixar, Toyota, and Analog Devices. The account is grounded in Amy Edmondson’s research on psychological safety, in Google’s internal Project Aristotle study, and in a 2017 meta-analysis of 136 independent samples that puts the link between safety and team performance on firm empirical footing. The pattern across all six is a matter of sequence, not softness. Encouragement comes first, scrutiny second, and the scrutiny, when it arrives, is exacting. lateralworks reaches the same conclusion from a different direction. Thirty-six years of field work on the fastest technology teams shows that innovation can be managed when leaders isolate the small slice of real invention, run it through fast learning cycles, and scrub it with a critical braintrust before it fails in production. The mechanism is identical: protect the fragile thing long enough to make it strong. We close with a practical model and a candid caveat. These six are famous survivors, and at least one of their programs eroded the moment leadership stopped defending it. The model works only while someone with authority keeps the space open.

The problem — How good ideas die

In most companies, innovation does not die for want of good ideas. It dies because the organization is hostile to ideas in their early, unfinished state. Skepticism, cost questions, and feasibility tests are useful tools. Applied too soon, they are lethal. When they arrive before an idea has had room to develop, two failures compound. Promising ideas get killed before they can be tested, and people learn that proposing something new is not worth the risk. The second failure is the expensive one. You lose the idea that was never developed, the person who would have developed it, and the culture of initiative that would have drawn in their replacement.

The paradox Why rigor turns hostile The paradox is that the discipline that makes an organization good at execution can make it deadly to invention. Execution rewards predictability and the removal of uncertainty. Innovation requires the opposite: exploration, tolerance for ambiguity, and a willingness to invest in an idea before its value is proven. Most organizations are tuned for the first and accidentally hostile to the second. The leader’s task is not to choose between rigor and creativity. It is to build an environment where both fit, where ideas get room to breathe before they meet the stress tests that will make them stronger or show that they should be set aside. The order of operations is the whole game. Figure 1. Two paths for the same raw idea. Scrutiny first kills the idea and, over time, the supply of ideas. Encouragement first protects the idea long enough to stress-test it and signals that thinking is valued, so the next idea surfaces. The research foundation Psychological safety The academic frame for this problem is Amy Edmondson’s work on psychological safety, which she defines as a shared belief that interpersonal risks—asking a question, admitting a mistake, voicing a half-formed idea, raising a concern—will not be punished. The term predates her: Schein and Bennis used it in 1965 to describe a group climate that reduces the fear of being judged, and Kahn revived it in 1990 in the context of engagement. The modern, testable version is Edmondson’s. Her discovery was counterintuitive. Studying medical teams in the 1990s, she expected the best teams to report fewer errors. They reported more. The better teams were not making more mistakes; they were surfacing and discussing them, while weaker teams hid them. Safety did not lower the bar. It raised the flow of information that lets a team learn.

Google reached the same conclusion from data. Between 2012 and 2014, Project Aristotle studied 180 teams across engineering and sales, coded more than 250 attributes, and looked for the recipe behind high performance. The strongest single predictor was not who was on the team. It was psychological safety, ahead of team composition, individual talent, and tenure. Four other dynamics completed the picture: dependability, structure and clarity, meaning, and impact. The effect is not an artifact of one study. A 2017 meta-analysis pooled 136 independent samples covering more than 22,000 individuals and nearly 5,000 groups, and confirmed a robust link between psychological safety and both task performance and citizenship behavior, over and above related factors such as leader relations and engagement. A systematic review the same year mapped the same nomological network across industries and cultures. The claim has the weight of evidence behind it. What it is not Edmondson is careful to separate psychological safety from niceness or the absence of accountability. A safe environment is not one where everyone agrees and mediocrity is tolerated. It is one where candor is normal enough that people can have the hard conversations that drive quality up. The leaders who build it well, she notes, are driven and demanding but have empathy. The target is what she calls the learning zone: high safety paired with high standards. Teams there take intelligent risks, share freely, and deliver. The dangerous opposite is the anxiety zone—high standards, low safety—where people fear punishment, stay quiet about concerns, and carry stress without learning. Many well-meaning, structurally misaligned organizations live there.

Figure 2. The learning zone. Psychological safety and accountability are two axes, not a trade-off. The goal is the high-safety, high-standards quadrant, where candor and rigor reinforce each other. After Edmondson. The 85 percent problem. Edmondson reports that 85 percent of employees have, at some point, withheld important information from a manager because they feared the consequences of speaking up. That is the cost of the anxiety zone, stated as a number: most of the information a leader needs is being filtered before it reaches them.

The evidence — Six companies that solved the paradox

Six companies, six mechanisms, one structure. Each built a way to give ideas room before subjecting them to the scrutiny that could kill them early. The mechanisms differ by industry and scale. The logic does not. A note on selection. These are famous companies chosen because their methods are well documented, which means they are also survivors. Read them for the structural pattern they share, not as proof that any single policy guarantees an outcome. Where a program struggled, this paper says so.

Case one and two Google and 3M: structured freedom Google’s 20% time In their 2004 founders’ letter, Larry Page and Sergey Brin told prospective shareholders that employees were encouraged to spend a fifth of their paid time on projects they believed would help the company. The idea was borrowed from 3M. The results were not marginal: AdSense, Google News, and the autocomplete behind Google Suggest all grew out of 20% time. Google News began as engineer Krishna Bharat’s personal project to improve news access after the attacks of September 11, 2001. The power was cultural, not logistical. By formalizing permission to explore outside a job description, Google signaled that it trusted employees to know where value might be found. That is decentralized innovation: every employee licensed to act as an inventor. Where it struggled The same case shows the limit. As Google grew and quarterly pressure mounted, many engineers could not actually find the time, and "120% time" became the wry internal name for doing it on top of a full load. Former HR chief Laszlo Bock described it as a value that waxed and waned with team dynamics and leadership support. Structured freedom produces extraordinary results, but only while leaders defend the space against delivery pressure. Without that defense, permission to innovate stays theoretical. 3M’s 15% culture 3M had been running its own version for decades, an ethos forged under the wartime pressure to invent or fall behind and rooted further back in William McKnight’s early insistence on autonomy. The most famous result is the Post-it Note. Spencer Silver discovered the low-tack adhesive in 1968; Art Fry found its use during his discretionary time in the early 1970s; the product reached the market in 1980. (The note is often misdated to the 1950s. It was not.) What set 3M apart was longevity. The 15% culture survived leadership changes and market shifts because it was treated as an operating principle, not a perk. A 3M maxim captures the whole problem this paper is about. As one of its R&D; managers put it, every great new product is killed at least three times by managers before it survives. The 15% culture exists to keep an idea alive through those three killings.

Case three Amazon: working backwards Most of Amazon’s major products since 2004 began with Working Backwards, and its core tool is the PR/FAQ: a mock press release plus a set of frequently asked questions, written before any development starts. The press release describes the finished product from the customer’s point of view. The FAQ then takes on the hard questions, both the customer’s and the internal business case. The design separates two acts that organizations usually jam together. Teams first imagine the most compelling result they can, dreaming big about what the product could be, before anyone raises cost, feasibility, or competitive risk. The stress test comes after the idea exists in its strongest form, not before it has one. Iteration without premature judgment A team may write ten or more drafts and meet senior leaders five or more times to argue and refine. Early drafts are explorations, not commitments. A team can learn through the process that an idea does not work, and that is counted as a success, because the discovery happened on paper rather than in production. Jeff Bezos reinforced the point with his two kinds of decision: Type 1 decisions are consequential and hard to reverse and deserve slow deliberation; Type 2 decisions are reversible and should be made fast. Treating most innovation bets as Type 2 lets teams act instead of waiting for a certainty that never comes.

Case four Pixar: the braintrust Pixar’s central mechanism, the Braintrust, is a recurring meeting of directors, writers, and storytellers who review films in progress and give candid feedback. Ed Catmull built it during Toy Story, when he saw that the film’s director had lost perspective inside the work, and described it in his book Creativity, Inc.. Two rules make it work. First, the Braintrust has no authority. It diagnoses problems; it does not prescribe fixes, and the director is free to ignore any specific suggestion. Removing the threat that feedback becomes a mandate is what makes it safe for a director to expose the weaknesses in their own work. Second, the feedback is candid but aimed at the work, never the person. Catmull described the job as pushing toward excellence and rooting out mediocrity. From bad to good Catmull was blunt that all Pixar films start out bad, and that the creative process is what carries them from there to great. Normalizing the idea that early work is supposed to be rough is itself a form of psychological safety: directors did not need to defend their films, they needed to expose the flaws so the group could help. The model travelled. When Disney bought Pixar and asked Catmull and John Lasseter to lead its long-struggling animation studio, they rebuilt the same conditions, and the turnaround produced Frozen, Big Hero 6, and a sustained run of hits. The Braintrust is not a film-industry quirk. It is a structural answer to the universal problem of getting honest input on fragile ideas.

Case five and six Toyota and Analog Devices: ideas at scale Toyota’s kaizen Kaizen, "change for the better," sits at the heart of the Toyota Production System. Toyota treats continuous improvement as a daily habit rather than a campaign run when something breaks. Its Creative Idea Suggestion System, launched in 1951, has run for more than seventy years. The striking figure is not the volume of ideas but the adoption rate. As the system matured, the share of suggestions actually implemented climbed from under a third in the mid-1950s to the mid-90s percent by the late 1980s, with employee participation around 95 percent. That is the engine of its longevity: people keep contributing because they see their ideas acted on. One example makes the philosophy concrete. A plant tour guide, tired of setting her bag on the floor during stops, suggested adding hooks to the catwalk. The change was tiny and had no measurable return. Toyota made it anyway, because the system values participation over magnitude. Reject the small ideas for want of obvious payoff and you lose the larger ones that would have followed. Analog Devices’ EngineerZone Analog Devices, a global semiconductor company, applied the same instinct to customers. In 2009 it launched EngineerZone, an online community that put design engineers in direct contact with the product developers behind ADI’s chips, replacing a one-to-one support model with a one-to-many community. ADI serves more than 60,000 core-market customers who have no direct sales support and rely on the web to self-serve, and the community gave them a route to the company’s own experts. Membership grew by more than 100 percent a year in its early period, and the community won industry recognition, including a 2013 Forrester Groundswell award for business-to-business social relationships. Innovation as a cultural shift The platform was the easy part. The hard part was convincing divisional leaders to let the engineers designing the next generation of products spend time answering customer questions in public. That took trust: that engineers would represent the company well, that open dialogue would strengthen rather than weaken relationships, and that the payoff would not always be immediately measurable. ADI also kept the community deliberately ad-free, resisting the pull to turn every touchpoint into a sales channel. That restraint was essential to its credibility. The parallel to lateralworks field practice is exact: the fastest teams make liberal use of outside experts because time to market is worth more than the risk of leaking an idea.

Figure 3. Six mechanisms, one structure. Each company built a different device, but every one protects an idea in its fragile state and teaches the team that contributing is safe and valued.

Pixar On candor and trust "Without the critical ingredient that is candor, there can be no trust. And without trust, creative collaboration is not possible." Ed Catmull, co-founder, Pixar Creativity, Inc. (2014)

The framework — A practical model for leaders

The six cases span industries, scales, and kinds of innovation. They share a structure that distills into five principles any leader can run. None of this requires a budget or a reorganization. It requires sequencing the work and protecting the early stage, which is a leadership behavior before it is a program.

The model Five principles 1. Sequence matters Encouragement precedes scrutiny. Google gave permission to explore before asking anyone to justify it. Amazon imagines the ideal result before questioning feasibility. Pixar reviews rough work rather than waiting for polish. Toyota implements first and measures later. In every case the stress test arrives after the idea has had room to develop. 2. Separate the idea from the evaluation Amazon’s press release separates imagining a great product from deciding whether to build it. Pixar’s Braintrust has no authority, so feedback is advice, not a verdict. The separation lets people advocate for an idea without feeling judged as people. lateralworks runs the same move through de Bono’s Challenge process: question the assumption, not the person who holds it. 3. Make feedback a ritual, not a surprise Pixar’s reviews are scheduled. Toyota’s suggestion review is continuous. Amazon’s PR/FAQ reviews are an expected part of the cycle. Predictable feedback builds safety, because people prepare differently for a known cadence than for an ambush. 4. Protect the space from above Every one of these programs needed active cover from senior leadership. Google’s 20% time eroded when leaders stopped defending it. ADI’s community survived because leaders kept marketing out. Pixar’s Braintrust held because Catmull personally maintained the norms of candor and non-authority. Innovation space does not maintain itself. 5. Reward the process, not only the outcome Toyota adopts most suggestions not because each is brilliant but because participation is the point. Pixar normalized rough early work. When people see that developing an idea is respected even when the idea pivots or dies, they come back with the next one.

Figure 4. The sequence and the five principles. Encouragement, development, stress-test, refinement, and reward run in order. The stress test is exacting, but it comes after the idea has room to develop, never before. The strategic case Innovation and retention There is a direct, under-appreciated link between innovation culture and whether people stay. When talented people feel their ideas are valued and that they have agency to improve their work, they stay. When new thinking is unwelcome and the safe move is to wait for instructions, they leave. The evidence runs the same direction. Project Aristotle found that teams with high psychological safety had lower turnover. Recent Harvard Business School research by Edmondson and colleagues finds that psychological safety guards against burnout and turnover precisely in hard times, which is exactly when cost pressure tempts leaders to cut it. For a leader worried about retention, this is not a soft argument. The cost of replacing a skilled person far exceeds the cost of an environment that keeps them, and the ideas they generate while engaged tend to dwarf the investment that produced them.

A Appendix The lateralworks parallel lateralworks arrived at the same destination by a different road. Thirty-six years of work on the fastest technology teams, from Sony’s first PlayStation to seven-billion-dollar semiconductor fabs, produced a method for managing innovation that mirrors the structure in this paper. The common objection on a hard program is that innovation cannot be managed, that ideas happen when they happen. lateralworks rejects that, having programmed and accelerated breakthroughs on bleeding-edge work, including sub-20-nanometer process development. Innovation cannot be scheduled to the day, but it can be managed. Being roughly right is better than being precisely wrong.

The parallel Managing innovation as learning Isolate the innovation Smith and Reinertsen observed that most of the work on a development project is known; usually less than ten percent is genuinely new. The discipline is to isolate that slice, resource it with the right experts, give it a wide window, and manage the rest conventionally. This is Amazon’s instinct to treat the new bet as a small, reversible experiment, expressed in schedule terms. Run fast learning cycles The truly new slice is managed with learning cycles—plan, do, check, act—and tracked against learning milestones rather than product-performance milestones. The principle is the one behind "fail fast": the faster the failure, the faster the learning. This is the same logic as Amazon counting a dead idea as a success when the discovery happens on paper, and Pixar expecting early films to be bad. Failure is an increment of learning, not a verdict on a person. Scrub it with a braintrust lateralworks adapted Pixar’s Braintrust and joined it to its own practice of critically scrubbing schedules and designs with small panels of external subject-matter experts. The panel’s job is to find the failure points, logic errors, and over-aggressive durations before they fail in production. As in Pixar’s version, the panel has no authority over the fix; the team converts the critique into improvement. Catmull’s line is the hinge: without candor there is no trust, and without trust there is no creative collaboration. The host creates the conditions The thread tying the paper to FTTM is the host. In lateralworks terms, teams deliver products; the host—senior leadership—creates the conditions under which they deliver fast. That is the same claim as principle four. The host protects the isolated slice, defends the cadence of review, and rewards the learning. Fix the host and speed becomes the default. Leave the host alone and the best team-level habits fade within a year.

Figure 5. The lateralworks parallel. Isolate the small slice of real innovation, run it through fast learning cycles, and scrub it with a critical braintrust, all under a host who protects the space. The same structure the six companies discovered, expressed as FTTM practice. What to do Monday. Pick one live idea that matters. Give it a protected window and a named sponsor with the authority to defend it. Schedule one candid review with no authority over the fix, aimed at the work and not the person. Treat the bet as reversible. Then do it again next week. Sequence and protection are behaviors, and they start with a single decision.

Sources

References

  1. Edmondson, A. C. "Psychological Safety and Learning Behavior in Work Teams." Administrative Science Quarterly, 44(2), 1999, pp. 350-383.
  2. Edmondson, A. C. The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley, 2019.
  3. Detert, J. R., and Edmondson, A. C. "Implicit Voice Theories: Taken-for-Granted Rules of Self-Censorship at Work." Academy of Management Journal, 54(3), 2011, pp. 461-488.
  4. Schein, E. H., and Bennis, W. G. Personal and Organizational Change Through Group Methods. Wiley, 1965.
  5. Kahn, W. A. "Psychological Conditions of Personal Engagement and Disengagement at Work." Academy of Management Journal, 33(4), 1990, pp. 692-724.
  6. Google re:Work. "Guide: Understand Team Effectiveness" (Project Aristotle). rework.withgoogle.com.
  7. Duhigg, C. "What Google Learned from Its Quest to Build the Perfect Team." The New York Times Magazine, February 25, 2016.
  8. Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., and Vracheva, V. "Psychological Safety: A Meta-Analytic Review and Extension." Personnel Psychology, 70(1), 2017, pp. 113-165.
  9. Newman, A., Donohue, R., and Eva, N. "Psychological Safety: A Systematic Review of the Literature." Human Resource Management Review, 27(3), 2017, pp. 521-535.
  10. Edmondson, A. C., Kerrissey, M., and Bahadurzada, H. Research on psychological safety, burnout, and turnover. Harvard Business School Working Knowledge, November 2025.
  11. Page, L., and Brin, S. "2004 Founders’ IPO Letter: An Owner’s Manual for Google’s Shareholders." Google, 2004.
  12. Bock, L. Work Rules! Insights from Inside Google That Will Transform How You Live and Lead. Twelve, 2015.
  13. Goetz, K. "How 3M Gave Everyone Days Off and Created an Innovation Dynamo." Fast Company, February 2011; Post-it Brand, "The History of the Post-it Note." post-it.com.
  14. Pink, D. H. Drive: The Surprising Truth About What Motivates Us. Riverhead, 2009.
  15. Bryar, C., and Carr, B. Working Backwards: Insights, Stories, and Secrets from Inside Amazon. St. Martin’s Press, 2021.
  16. Bezos, J. "2015 Letter to Shareholders." Amazon, April 2016.
  17. Catmull, E., and Wallace, A. Creativity, Inc.: Overcoming the Unseen Forces That Stand in the Way of True Inspiration. Random House, 2014.
  18. Catmull, E. "How Pixar Fosters Collective Creativity." Harvard Business Review, September 2008.
  19. Liker, J. K. The Toyota Way: 14 Management Principles from the World’s Greatest Manufacturer. McGraw-Hill, 2004.
  20. Ohno, T. Toyota Production System: Beyond Large-Scale Production. Productivity Press, 1988.
  21. Yasuda, Y. 40 Years, 20 Million Ideas: The Toyota Suggestion System. Productivity Press, 1991.
  22. Analog Devices, Inc. "Analog Devices Expands Online Technical Support Community for Design Engineers." BusinessWire, February 5, 2013.
  23. DiMauro, V., with Mitchell, J. "Spotlight on Analog Devices’ Award-Winning B2B Online Community: EngineerZone." Social Media Today / CustomerThink, 2013.
  24. de Bono, E. Lateral Thinking: Creativity Step by Step. Harper & Row, 1970.
  25. Reinertsen, D. G. The Principles of Product Development Flow: Second Generation Lean Product Development. Celeritas, 2009.
  26. Smith, P. G., and Reinertsen, D. G. Developing Products in Half the Time. Van Nostrand Reinhold, 1991.
  27. Wheelwright, S. C., and Clark, K. B. Revolutionizing Product Development: Quantum Leaps in Speed, Efficiency, and Quality. Free Press, 1992.
  28. lateralworks. "Braintrust," "Learning Cycles," "Innovation Can’t Be Predicted," and "FTTM Innovation." Ideas library, lateralworks.com, 2019-2026.
  29. lateralworks. Internal assessment database. Engagement data across client programs, 1988-2026.