101:28
explore vs exploit
Osterwalder argues that most corporate innovation failure is a systems and governance problem, not a talent or money problem: organizations default to applying the KPIs, business plans, and single-mindset management built for exploiting known businesses to the fundamentally different, high-uncertainty work of exploring new ones, and fixing this requires a dual-portfolio system with staged funding, phase-appropriate evidence requirements, and empowered sponsorship for scaling validated ideas.
Companies must run two coexisting modes — "explore" (searching for new business models, high uncertainty) and "exploit" (managing known ones, low uncertainty) — and most innovation failure comes from applying exploit-style KPIs, planning, and business plans to explore-stage work.
Business plans and spreadsheets function as "fantasy made explicit" when built without field evidence, and can lock organizations into executing something nobody wants.
Reducing innovation risk is only achieved through evidence gathered from the real world (talking to people, testing, experimenting), not through more analysis alone.
Companies should manage a dual portfolio: an Exploit Portfolio (existing businesses, judged on return vs. disruption risk, managed via acquisition/improvement/divestiture) and an Explore Portfolio (new bets, judged on expected return vs. innovation risk, managed by funding many small experiments and killing most of them).
Bosch's internal accelerator (est. 2015) is presented as a working case: staged funding (~€120K then ~€300K per team), no building allowed in phase one, team self-assessed go/no-go, and an overall ~10% success rate.
A product needs a viable business model to succeed — a well-loved product can still fail financially, and some value propositions only work with a different business model (Xerox 914's lease-plus-per-copy model, Apple's iPod deal with record labels).
Evidence exists on a spectrum from stated opinion (weakest) through tangible reactions, lightweight signups, click-throughs, and financial or reputational commitment, up to real market evidence (strongest); bigger investment decisions should require stronger evidence.
Innovation capability is diagnosed via three factors: how much time leadership spends on innovation, whether an empowered innovation leader reports to the CEO, and whether the organization can kill 9 out of 10 ideas — framed as "power problems," not money problems.
Killing ideas early and often is essential; companies that can't kill ideas end up with "zombie projects" that consume resources without justification.
Scaling a validated idea requires an organizational "landing pad" (a receiving business unit or even an entirely new P&L, as with Amazon Web Services) — without one, good ideas go homeless or get stripped for parts and killed.
Strategyzer applied its own explore/exploit and evidence-testing discipline to its own business model, deliberately self-disrupting a consulting-fee model into a licensed "programs on platforms" model.
Growth boards reviewing innovation decisions should be staffed with people who hold real organizational power to remove blockers, with evidence requirements that scale up by project phase rather than demanding full evidence too early.
Hackathons and idea competitions are "innovation theater" unless connected to a defined system that captures and channels their outputs into the broader innovation pipeline.
Ping An, contrasted with Bosch, is presented as a company that integrated corporate venture capital, M&A, open innovation, and an internal innovation team under one strategy, rising from roughly a top-500 company to about top 20–30 globally.
Business Model Canvas — A nine-block visual tool, originally called the "business model ontology" in Osterwalder's PhD, for mapping how an organization creates, delivers, and captures value. Apply: Score evidence separately for each of its nine blocks (customer jobs/pains/gains, value proposition, feasibility, viability, adaptability) before deciding whether to scale an idea.
Value Proposition Canvas / Value Proposition Design — A companion tool to the Business Model Canvas that maps customer jobs, pains, and gains against a product's pain relievers and gain creators. Apply: Apply the jobs/pains/gains lens to any stakeholder — senior leaders, team members, even authors — before trying to introduce change, not only to end customers.
Testing Business Ideas methodology — A framework from Osterwalder and David Bland's book for scoring the strength of evidence behind a business idea instead of treating all evidence as equal. Apply: Rate the strength of evidence gathered for each Business Model Canvas block rather than relying on a single willingness-to-pay data point.
Explore vs. Exploit framework — A distinction between "search" mode (finding a new, high-uncertainty business model) and "exploit" mode (managing and scaling a known, low-uncertainty one). Apply: Switch the questions asked and KPIs applied depending on which mode a given project is actually in, rather than defaulting to exploit-style oversight for exploratory work.
Ambidextrous organization — The academic term for an organization that runs explore and exploit simultaneously and well. Apply: Build separate but connected governance and teams for exploration and exploitation instead of managing both with one system.
Dual-portfolio management (Exploit + Explore Portfolios) — Managing two parallel portfolios: existing businesses judged on return vs. disruption risk, and new bets judged on expected return vs. innovation risk. Apply: For exploit, acquire/partner, improve, or divest business models; for explore, ideate and test many small bets and only transfer proven ones into the exploit portfolio.
Innovation funnel with early-kill discipline — A staged pipeline of many small-bet ideas in which most are deliberately killed early because winners can't be identified in advance. Apply: Fund a large number of small experiments rather than a few large ones, and cut non-performers early to free resources for survivors.
Bosch's staged accelerator (Competence Center for Business Model Innovation) — A two-phase internal accelerator, set up in 2015, with phase one (~3 months, ~€120K per team, no building, focused on customer jobs/pains/gains evidence) and phase two (~€300K per team, MVP building), team self-assessed go/no-go, and an overall ~10% success rate. Apply: Replicate the staged-budget, staged-evidence structure: require desirability evidence before funding any building, and let teams recommend their own continue/kill decision.
Evidence spectrum (ladder of evidence strength) — A hierarchy running from stated opinion (weakest) through tangible mashup reactions, lightweight signups, click-throughs, and financial/reputational commitment, up to real market evidence (strongest). Apply: Match the tier of evidence demanded to the size of the investment decision, and avoid basing large bets on what people merely say.
URL test — A technique of sending a specific call-to-action link to check whether stated interest converts into an actual click. Apply: Use it to catch stakeholders (e.g., senior leaders) who claim interest but won't act, as a cheap check of the say-do gap.
Say-do gap / directional evidence — The principle that stated intent ("say") and actual behavior ("do") are different strengths of evidence, with paid or behavioral commitment being stronger. Apply: Seek pre-orders, down payments, or other paid commitments rather than survey responses before committing larger investment.
Feasibility / Viability / Desirability / Adaptability risk breakdown — Four risk categories used to structure evidence-gathering: desirability (customer jobs/pains/gains and value-prop fit), feasibility, viability (financial), and adaptability (external context like regulation and competition). Apply: Score evidence separately for each category instead of treating "risk" as a single undifferentiated variable.
Growth board with phase-gated evidence requirements — A governance body that reviews innovation projects against evidence thresholds set for each phase (discovery → validation → acceleration/pre-scaling → scale). Apply: Staff it with people who hold real power to remove organizational blockers, and set explicit, escalating evidence bars per phase instead of demanding full evidence too early.
Three-factor innovation-capability diagnostic — A check of whether a company can innovate at scale based on (1) leadership time spent on innovation, (2) an empowered innovation leader reporting to the CEO, and (3) willingness to kill 9 of 10 ideas. Apply: Use it to diagnose why an organization isn't innovating, treating shortfalls as power/attention problems rather than money problems.
Discovery-sprint approach for organizations lacking leadership buy-in — Running 5–10 small projects through the innovation funnel over roughly three months to surface organizational blockers as evidence, even without full management buy-in. Apply: Use the resulting evidence of what stopped teams as the pitch to leadership for building a larger, formal innovation system.
"Landing pad" for scaling ideas — A deliberate organizational mechanism or sponsorship for absorbing a validated idea into (or alongside) the core business, addressing the failure mode where "homeless ideas remain homeless.". Apply: Designate in advance where a successful idea will land — an existing business unit or a new P&L, as with AWS at Amazon — before scaling it.
Steve Blank's Creator vs. Entrepreneur distinction — A distinction between people who repeatedly start new ventures without ever scaling one ("Creator") and those who scale a venture ("Entrepreneur"). Apply: Use it for self-diagnosis of founder or leadership patterns to recognize when a person or team needs to shift from starting new things to scaling one.
Steve Blank's "no titles" team-size heuristic — A heuristic that at very small team sizes (3, 5, 10 people), formal titles are the last thing a startup needs and focus should stay on reaching product-market fit. Apply: Delay introducing formal roles and structure in an early-stage team until the messiness of finding fit is resolved enough to warrant it.
MVP-as-mashup ("mule"/"Frankenstein" prototype) — Building a minimum viable prototype by combining existing components rather than engineering from scratch, exemplified by Tesla's early "mule" built from a Lotus Elise chassis plus a battery and drivetrain. Apply: Combine off-the-shelf parts or tools into a rough test bed to learn quickly and cheaply before committing to bespoke engineering.
10-second house-drawing exercise — A live exercise where participants draw a house in five seconds, then someone else adds to it in another five seconds, exposing shared mental models and default assumptions. Apply: Use it in a workshop to demonstrate that people default to conventional mental models about products and business models unless deliberately pushed to think differently.
Self-disruption / unbundling of one's own business model — Strategyzer's own case of deliberately shifting from a consulting-fee model to a licensed "programs on platforms" model, eventually telling clients they no longer need the company as consultants. Apply: Apply the same test-and-iterate discipline to your own business model, not just your product, and be willing to cannibalize an existing revenue stream.
"Business model fit" (beyond product-market fit) — The idea that startups need to find a viable business model in addition to product-market fit. Apply: After validating that customers want the product, separately validate that the surrounding business model (pricing, distribution, cost structure) can make more money than it spends.
Marshall Goldsmith's "let it go" principle — Guidance, attributed to leadership coach Marshall Goldsmith (who coached Alan Mulally through Ford's turnaround), that when a decision is outside your decision-making power, you should let it go. Apply: After attempting to reframe stakeholders' thinking, if a decision remains outside your authority, disengage rather than keep fighting it, and consider whether it's the right role for you.
Portfolio-of-connected-programs model — An integration model linking leadership evidence-based-investment programs, hackathons, and a defined follow-on "vehicle," contrasted with running corporate VC, M&A, open innovation, and an internal innovation team as disconnected silos. Apply: Connect hackathon outputs and other idea sources to a defined pipeline and landing pad, and unify strategy across CVC, M&A, open innovation, and the innovation team instead of letting each operate independently.
Analysis time scales the wrong way with company size: startups tend to under-analyze under time pressure, while large companies over-analyze because they have time to spare — Osterwalder claims a week of large-company analysis often only needs about four hours.
When Bosch made evidence criteria transparent in advance, team self-assessments of go/no-go matched management's actual decisions about 90% of the time, suggesting transparent evidence bars — not management judgment — drive alignment.
More funding can increase the risk of innovation failure rather than reduce it: Better Place burned roughly $850 million and Quibi burned $1.3 billion of a $1.7 billion raise in four months, both cited as cases where large upfront capital let flawed ideas scale before being tested.
A single strong value proposition doesn't guarantee success: Osterwalder directly rejects a team member's claim that the value proposition is the "most important" Business Model Canvas block, insisting all nine blocks carry equal weight.
"Innovators will innovate" once organizational blockers are removed — Osterwalder says he initially believed extra incentives were required to spur bottom-up innovation, but changed his view to see the constraint as structural (walls/blockers), not motivational.
Post-success incentive design for internal innovators who build a large new business inside a company is described as an unsolved problem, with Cisco's attempt at addressing it cited as not really working.
The core scaling failure mode isn't only picking the wrong ideas too early — it's equally often a company's inability to give a validated idea a home ("landing pad"), which is why he says Amazon Web Services required an entirely new P&L rather than fitting into an existing business line.
Osterwalder says he deliberately avoids jargon like "Lean Startup" or "testing" when introducing structural change to senior leaders unfamiliar with it, instead applying jobs/pains/gains thinking to the stakeholders themselves, not just to customers.
Formally handing a validated idea over from a "testing team" to an "executing team" doesn't work well in his experience, which he compares to how startup CEOs almost never voluntarily declare their own founder phase over.
«What Richard didn't say is I can be a pain in the ass when we work on a book.»
— 03:53
«Innovation is bottom up but without the top down support and systems doesn't actually work.»
— 05:33
«It's because the system is broken.»
— 08:08
«If you apply the KPIs, the key performance indicators, culture and process of this world to this world, you just killed Innovation.»
— 09:48
«Business plans and business cases are the death penalty of innovation.»
— 12:24
«If you don't have the data you can analyze as much as you want — it's Fantasy Land.»
— 15:05
«Even the best business models expire like a yogurt in the fridge.»
— 21:35
«I like business models because at the end of the day we're not building products, we're building businesses or growth engines.»
— 23:10
«Innovation is the crappiest industry you can be in because there's no money there's no power»
— 50:11
«spreadsheets are basically a fantasy made explicit»
— 55:37
«for every one of our failures we had a spreadsheet that looked awesome»
— 56:24
«what people say and what people do is not the same strength of evidence»
— 65:43
«No evidence is a business plan — a spreadsheet, a fantasy made explicit.»
— 72:16
«This is rarely a money problem, it's a power problem.»
— 78:41
«homeless ideas remain homeless»
— 96:54
«I used to think hackathons are the purest form of innovation theater»
— 99:19
Reception
Comments are sparse but generally appreciative, with one viewer even valuing the channel's low visibility as a personal edge.
Osterwalder builds a systems-level argument, anchored in named client cases (Bosch, Ping An, and Strategyzer's own business-model pivot) and audience-tested quiz questions, that corporate innovation failure is chiefly a governance and evidence-discipline problem rather than a money or talent problem; the talk functions as a live teaching session for his own frameworks (Business Model Canvas, the evidence spectrum, dual-portfolio management) rather than a neutral survey of competing views on innovation management.
101:28