Exploratory Responses: Action Orientation, Experimentation
ARC Phase 2
Image created with MidJourney, Paul Epping
Sailing into the Set: Exploratory Responses through Experimentation
Once you’ve reframed uncertainty into context—scanning the horizon like a surfer reading the swell patterns—the ARC propels you forward into exploratory responses. This is where the rubber meets the reef, or rather, where your board meets the water. No more frozen analysis from the shore; it’s time to sail out, drop in, and experiment. In the tech-surfing world, experimentation isn’t reckless charging; it’s deliberate, playful probing of the waves. You’re not committing to a full barrel ride yet—just testing the current, feeling the push and pull. This step breaks the fear-response paralysis by turning unknowns into actionable insights, fostering a mental model of curiosity over caution. As we’ll explore, core practices include safe-to-fail experiments, a portfolio approach, and a learning orientation. But to ride this wave fully, we’ll add iterative feedback loops, collaborative diversity, and scalable prototyping—tools that amplify your adaptive edge in business and personal tech swells.
Each small wave is an experiment. Each ride generates data. The surfer adjusts based on what they learn: “The waves are faster than they looked—I need to take off earlier.” “The current is stronger than I thought—I’ll position further north.” “This board feels too big for these conditions—I’ll switch to the smaller one.”
By the time a set of larger waves arrives, the surfer has run multiple experiments. They’ve reduced uncertainty through action, not analysis. They know the conditions not theoretically but experientially. They’re ready to commit to the bigger opportunity because they’ve built capability through progressive experimentation.
During the time I was involved in large transformations of companies, our approach was mainly based on the insights gained from exponential organizations. (Ismail, et all, 2014, 2023). Of the eleven attributes that have been identified in exponential organizations, one of the most powerful, and my favorite, turned out to be experimentation.
Core Point 1: Safe-to-Fail Experiments
Organizations trapped in the fear response cycle avoid experiments because experiments might fail, and failure feels threatening. But in complex, uncertain contexts—which technological disruption always creates—experimentation is the only reliable way to learn.
The Cynefin framework distinguishes between complicated contexts—where analysis and expertise can reveal the right answer—and complex contexts—where the right approach emerges only through experimentation and adaptation (Snowden & Boone, 2007). Technological disruption belongs to the complex domain: cause-and-effect relationships exist, but they can only be discerned in retrospect. You cannot predict precisely how AI will reshape your industry; you can only probe, sense, and respond.
The key is making experiments safe to fail—designed so that failure provides learning without threatening organizational survival.
For this reason, the following principles may be helpful. I’ve used these principles during transformations of companies and have been proven to be helpful.
Safe-to-fail design principles:
Contained scope: Test with one team, one product line, one geography—not the entire organization simultaneously. When Spotify experiments with new features, they release to 5% of users first, then 20%, then 50%, scaling only if results warrant (Kniberg & Ivarsson, 2012).
Time-boxed: “We’ll run this experiment for 30, 60, or 90 days, then evaluate.” The time frame depends on the complexity of the experiments, e.g., how many dependent processes are involved. Time limits prevent experiments from consuming resources indefinitely while preventing premature abandonment before learning occurs.
Clear hypotheses: Probably the most important one, but hugely underestimated. Reasoning like: “We believe that [X customers] will [Y behavior] because [Z reason].” Explicit hypotheses make learning possible. When the experiment ends, you know what you learned, not just whether it “worked.”
Defined success metrics: Not “Did it succeed?” but “Did we see [specific measurable outcomes]?” This separates learning (always successful if you gain insight) from commercial success (which may take multiple iterations).
End-user involvement: It operationalizes the “sense” portion of “probe, sense, respond” and ensures experiments generate actionable knowledge rather than just internal validation. Without it, organizations risk building the wrong thing, safely—technically sound solutions that fail to address actual needs.
Kill criteria: “We’ll stop if we see [specific warning signs].” Pre-defined exit criteria prevent sunk cost fallacy—continuing because you’ve already invested, not because the path is promising.
Amazon’s approach exemplifies this discipline, and they expect most experiments to fail: “If you’re going to take bold bets, they’re going to be experiments. And if they’re experiments, you don’t know ahead of time if they’re going to work” (Stone, 2013). Amazon Web Services started as a safe-to-fail experiment: a small team, limited resources, a clear hypothesis (”developers will pay for infrastructure as a service”), and defined metrics. It could have failed without threatening Amazon’s retail business. Instead, it became a multi-billion-dollar business.
The surfer’s wisdom: The experienced surfer doesn’t commit to the biggest wave of the day on their first ride. They test the conditions on smaller waves first—safe-to-fail experiments that build understanding without catastrophic consequences if they wipe out.
In surfing, no one designs a board alone in a lab. They prototype, then hand it to surfers in actual breaks to reveal what works. Technological experiments need the same real-world feedback loop—not just internal testing, but end-user co-creation from day one.
Core Point 2: Portfolio Approach
A single experiment tells you little. It might succeed because your hypothesis was right, or because you got lucky. It might fail because your hypothesis was wrong, or because execution was poor. You need a portfolio of experiments to distinguish signal from noise and to ensure you’re exploring multiple possible futures simultaneously.
Research on innovation portfolios demonstrates that successful companies maintain balanced portfolios across three horizons (Nagji & Tuff, 2012):
Horizon 1: Core business optimization (70% of resources)
- Incremental improvements to existing products and processes
- High probability of success, modest returns
- Timeline: Quarters
- Example: Using AI to improve existing customer service response times by 20%
Horizon 2: Adjacent expansion (20% of resources)
- Extensions into related markets or capabilities
- Moderate probability of success, significant returns
Timeline: 6 months -1 year
Example: Launching an AI-powered product recommendation engine that opens a new revenue stream
Horizon 3: Transformational bets, edge developments (10% of resources)
- Entirely new business models or markets
- Low probability of success, potentially massive returns
Timeline: 3-5 years
Example: Building AI coordination platform that could replace the current business model (Choudary,2025)
This portfolio approach ensures you’re simultaneously optimizing today’s business (H1), building tomorrow’s business (H2), and exploring the day-after-tomorrow’s business (H3). The fear response cycle causes organizations to over-invest in H1 (protecting the core) and under-invest in H2 and H3 (exploring the future).
Portfolio discipline prevents two common errors:
Error 1: Betting everything on one big experiment —If it fails, you’ve learned little and wasted significant resources. If it succeeds, you’ve learned only that this specific approach works, not why or what else might work.
Error 2: Running only tiny experiments —You never commit enough to learn whether something could work at scale. You’re experimenting, but not learning. Think big, act small. Instead of your city or region, the universe is your playground.
The portfolio balances these extremes: multiple experiments at different scales, with different risk/return profiles, testing various hypotheses about the evolving context.
The surfer’s wisdom: The experienced surfer doesn’t just watch one section of the break. They scan the entire lineup—inside, outside, left peak, right peak—maintaining awareness of multiple possible opportunities. When a set arrives, they’re positioned to catch whichever wave offers the best ride, not locked into a single option.
Core Point 3: Learning Orientation
The purpose of experiments isn’t just to succeed—it’s to learn. This distinction is crucial. Organizations with performance orientation celebrate only successes and hide failures. Organizations with a learning orientation extract insight from both and encourage sharing failures.
Among many studies about learning organizations, and for this book, I’m focusing on the seminal work of Amy Edmonson.
Intelligent failure framework (Edmondson, 2023):
Preventable failures: Mistakes from inattention, process violations, or known risks ignored
- Response: Fix the process, improve training, increase accountability
Example: Launching an experiment without defining success metrics
Complex failures: Failures from novel combinations or unforeseen interactions in complex systems
- Response: Analyze deeply, share learnings, update mental models
Example: An AI system produces unexpected results due to data interactions that no one anticipated
Intelligent failures: Failures from well-designed experiments in uncertain territory
- Response: Celebrate the learning, apply insights to next experiments
Example: Hypothesis that customers would value feature X proved wrong, but revealed they actually need feature Y
Only intelligent failures deserve celebration, but all three types deserve analysis. The learning orientation means treating experiments as learning investments, not success/failure binaries.
Practice: The 48-hour post-mortem
Within 48 hours of any significant experiment concluding (whether it “succeeded” or “failed”), conduct a structured review:
1. What did we predict? (Review original hypothesis)
2. What actually happened? (Objective results, data))
3. Why the gap? (Analysis of difference between prediction and reality)
4. What did we learn about the context? (Insights about technology, customers, competitors, our capabilities)
5. What should we do differently next time? (Applied learning)
6. Who else should know this? (Knowledge sharing)
This ritual transforms experience into organizational learning. Without it, experiments generate activity but not insight.
The surfer’s wisdom: After every session, experienced surfers reflect: “What worked? What didn’t? What did I learn about these conditions? What will I try differently next time?” They’re building a mental library of patterns that makes them progressively more skilled at reading and responding to varied contexts. During my active period as a surfer, we sat together and shared experiences. That was a natural process. I witness the same behavior when I’m visiting surfing places in Portugal. After being out and about on the water, there is time for reflection when they return to base and have a drink with their mates.
Additional Point: Rapid Iteration Cycles
Beyond safe-to-fail design, portfolio approach, and learning orientation, adaptive organizations master rapid iteration—completing experiment cycles faster than competitors.
Eric Ries’s Lean Startup methodology emphasizes the “build-measure-learn” loop: build a minimum viable product, measure how customers respond, learn what to build next (Ries, 2011). The faster you complete this loop, the more learning cycles you achieve in the same time period.
Speed compounds: If your experiment cycle takes six months and competitors’ takes four months, they complete twice as many learning cycles in the same period. After one year, they’ve run four experiments,
and you two. After two years, eight and you (only) four. The learning gap compounds exponentially.
Accelerating iteration:
- Reduce scope: Build the minimum that tests the hypothesis, not the complete vision. Spotify’s “think it, build it, ship it, tweak it” approach completes cycles in weeks, not months (Kniberg & Ivarsson, 2012).
Parallel experiments: Run multiple experiments simultaneously rather than sequentially. Amazon famously runs thousands of A/B tests concurrently.
Automated measurement: Build instrumentation into experiments from day one so results are visible in real-time, not after manual analysis weeks later.
Pre-defined decision rules: “If we see X, we do Y” eliminates decision delays between learning and action.
Keep rejected experiments: Document not just successful experiments but also rejected ones—including why they were rejected, what was learned, and under what conditions they might become viable. Archive not just what you built and tested, but what users told you.
Script for keeping information about rejected experiments
Experiment title: [Descriptive name]
Date: [Start - End]
Team: [Who led it, who participated]
Status: [Killed / Paused / Pivoted / Scaled]
Hypothesis: We believed that [X customers] would [Y behavior] because [Z reason]
Experiment design: - What we built/tested: [Description] - Who we tested with: [User segment, sample size] - Duration: [Timeline] - Success metrics: [What we measured]
Results: - Quantitative: [Data] - Qualitative: [User feedback, observations]
Analysis: Why it succeeded/failed: [Root cause] - Was the hypothesis wrong? - Was the execution flawed? - Was the timing off? - Was something else the issue?
Decision & Rationale: We decided to [kill/pause/pivot/scale] because [reasoning]
Key learnings: - About customers: [What we learned] - About technology: [What we learned] - About our capabilities: [What we learned] - About the market: [What we learned]
Revival conditions: This might work if: - [Condition 1: e.g., “Cloud infrastructure costs drop by 50%”] - [Condition 2: e.g., “Regulatory environment changes to allow X”] - [Condition 3: e.g., “Customer segment becomes more tech-savvy”] Signals to watch: - [What would indicate conditions are changing?]
Related experiments: - [Links to similar experiments, predecessor experiments, successor experiments]
Contact: - [Names of people with deep knowledge of this experiment]
Artifacts: - [Links to: prototype, data, presentations, user research, technical documentation]
The organization that iterates faster learns faster. The organization that learns faster adapts faster. The organization that adapts faster wins.
The surfer’s wisdom: Experienced surfers remember waves they didn’t catch—not with regret, but with analysis. “That wave looked good but closed out too fast—I was right not to go for it.” “That wave I passed on actually would have been perfect—I misjudged it.” They archive these observations mentally, building a library of pattern recognition that makes them better at reading future waves. They don’t forget their “failures” to catch waves; they learn from them. And sometimes, a wave pattern they rejected in the morning becomes the perfect wave in the afternoon when the tide shifts. The context changed; the opportunity returned.
Reflection Box
Exploration isn’t endless; it’s purposeful probing that feeds into decisions. But beware the trap of “analysis paralysis disguised as experimentation”—too many tests without synthesis can leave you treading water. We’ll address balancing this in later swells.
Outcome: Reduced Uncertainty Through Action
Exploratory responses—safe-to-fail experiments, portfolio approach, learning orientation, rapid iteration—don’t eliminate uncertainty. They reduce uncertainty to actionable levels while building organizational capability.
After Stage 1 (Context Perception), you’ve reframed uncertainty as navigable context. After Stage 2 (Exploratory Responses), you’ve actually navigated some of that context. You have experiential knowledge, not just theoretical understanding. You know what works in your specific situation, not just what worked for others.
This experiential learning creates confidence later on (Stage 8) that feeds back into better context perception (Stage 1) in the next cycle. The ARC is now in motion.
The surfer who has run experiments on smaller waves approaches the larger set with genuine confidence—not bravado, but earned capability. They’ve tested the conditions. They know their equipment. They’ve built the muscle memory. They’re ready to commit when the opportunity arrives.
That commitment—knowing when and how to move from exploration to execution—is Stage 3: Right -Speed Decisions - Accelerate.
References
Edmondson, A. C. (2023). *Right kind of wrong: The science of failing well*. Atria Books.
Kniberg, H., & Ivarsson, A. (2012). *Scaling agile @ Spotify with tribes, squads, chapters & guilds*. Spotify Technology S.A.
Nagji, B., & Tuff, G. (2012). Managing your innovation portfolio. *Harvard Business Review*, 90(5), 66-74.
Ries, E. (2011). *The lean startup: How today’s entrepreneurs use continuous innovation to create radically successful businesses*. Crown Business.
Snowden, D. J., & Boone, M. E. (2007). A leader’s framework for decision making. *Harvard Business Review*, 85(11), 68-76.
Stone, B. (2013). *The everything store: Jeff Bezos and the age of Amazon*. Little, Brown and Company.
Ismail, S., Diamandis, P., Malone, S. (2023). Exponential Organizations 2.0. The New Playbook for 10x Growth and Impact. Ethos Collective
Hunt, V., et al. (2018). Delivering Through Diversity. McKinsey & Company.
Ries, E. (2011). The Lean Startup. Crown Business.
Taleb, N. N. (2012). Antifragile: Things That Gain from Disorder. Random House.