Learning from Surprising Events
Pattern: A named solution to a recurring problem.
Use a small, mixed team to explain the gap between an expected outcome and a significant actual outcome, expose the assumption behind the gap, and change policy accordingly.
Also known as: Systematic Learning from Significant Surprising Events; Learning from Experience; Learning from ONE Event.
A surprise is evidence that the organization’s working model of reality is incomplete. The immediate response may contain the event, celebrate it, or assign fault. None of those responses necessarily explains why reasonable expectations failed. Learning from Surprising Events turns one consequential exception, good or bad, into a disciplined search for the assumption that no longer held.
Understand This First
• Assumption — the belief or condition under a cause-and-effect claim.
• Categories of Legitimate Reservations — the scrutiny used to challenge candidate explanations.
• Future Reality Tree — the tool for testing the policy changes that follow from the revised understanding.
Context
TOC has established routes for familiar problem shapes. A Current Reality Tree connects chronic undesirable effects to a small number of root causes. Daily Management with TOC uses a Cloud, a Negative Branch Reservation, and an IO Map to turn daily conflicts and firefighting into better operating rules.
Eli Schragenheim argues that a significant one-time surprise has a different starting condition. The team doesn’t yet recognize the cause, so it can’t reliably translate the event into an obvious operational conflict. Its expectation had looked sound until reality contradicted it. A major customer leaves despite a strong relationship. A product succeeds for reasons the company didn’t plan. A supposedly routine action produces an unprecedented failure. The learning problem is the gap itself.
Schragenheim and Avner Passal developed a structured method for that gap. The 2022 TOCICO-recognized paper presents the current eight-step form. Earlier presentations described a seven-step form that used the gap as the opposing actions of an Evaporating Cloud. The recognized version doesn’t build a Cloud at the operational cause because it begins before an obvious conflict has been found.
Problem
Organizations learn poorly from exceptions. Under pressure, people settle on the first plausible explanation, usually one that fits what they already believe. A failure becomes evidence that someone was careless. An unexpected success confirms the favored strategy. The story closes before competing explanations have been stated or tested.
Blame makes this worse because it narrows the inquiry to the person nearest the event. Schragenheim’s position is that blame usually adds no positive value to learning, though he preserves an exception for clear negligence or bad intentions. Removing casual blame doesn’t remove accountability. It keeps the team looking for the assumptions, policies, conditions, and interactions that made the outcome possible.
The opposite failure is endless analysis. A team can produce many possible causes and still have no basis for choosing among them. A useful method must expand the hypothesis set before it narrows it, distinguish invalidation from proof, and end with changed behavior rather than an interesting story.
Forces
• The event is emotionally loaded. Failure invites defense and blame; success invites self-congratulation.
• The expectation was reasonable before the event. Simply reversing the old assumption throws away the cases in which it worked.
• Several explanations can fit the same observation. A plausible causal story isn’t proof that the story caused this outcome.
• Insiders know the event but share its assumptions. Outsiders see different possibilities but may miss operational facts.
• Learning matters only if policy changes. A corrected explanation that leaves norms and decisions untouched won’t improve the next outcome.
Solution
Run the eight-step process with a team of three to five people. Include people who were involved in the event and outsiders who can challenge the insiders’ shared model.
1. Choose a significant event. Select an outcome that departed materially from expectation and whose explanation could improve future decisions. A surprise can be favorable or unfavorable.
2. Create the team. Mix firsthand knowledge with distance. Keep the group small enough to investigate one event together.
3. Verbalize the gap. State what the team expected, what actually happened, and why the difference matters. Schragenheim reports that this step often took longer than expected; once the gap was clear, later work moved faster.
4. List every plausible explanation. Delay evaluation. Use additional-cause thinking to resist the urge to stop at the first satisfactory story.
5. Detail and test the explanations. Spell out the causal chain each hypothesis requires, then compare it with the observed event. Use effect-cause-effect reasoning to invalidate explanations whose expected traces aren’t present.
6. Identify the flawed paradigm or assumption. Work down from the operational explanation to the belief that supported the mistaken expectation.
7. Develop changes in policies and norms. Don’t merely reverse the assumption. Record earlier cases where it worked and identify what differed here. Use a Future Reality Tree and Negative Branch Reservation when the proposed changes need causal and side-effect testing.
8. Share the new understanding. Tell the event as a learning story: expectation, surprise, hypotheses, invalidations, revised assumption, and changed rule. The reach may extend beyond the team that experienced the event.
Invalidation is not certainty: Schragenheim treats effect-cause-effect as stronger for invalidating a hypothesis than for proving one. Even proof by eliminating alternatives is only partial when the team can’t know that it listed every possible cause. Record the confidence and the unresolved alternatives.
How It Plays Out
A lost customer exposes a hidden condition. Schragenheim illustrates a company losing a major customer despite years of apparent satisfaction. “The relationship deteriorated” is too broad, and “the account manager failed” closes the inquiry too early. The team states what it expected, lists possible changes in the customer’s environment and in its own behavior, and asks what evidence each explanation predicts. The useful result isn’t a culprit. It is a more precise assumption about what maintains the relationship and a policy for detecting when its supporting conditions change.
An unexpected success deserves the same rigor. A soup that sells much better than forecast can be dismissed as luck or credited to the most visible change. The team instead tests ingredient, presentation, timing, customer, and context explanations. If the evidence supports a condition the recipe team hadn’t considered, the organization can preserve or reproduce that condition. A favorable surprise can reveal a flawed paradigm just as clearly as a failure can.
These are illustrations, not measured case results. The 2022 paper reports that the process was used in several security-oriented organizations in Israel, including organizations that weren’t TOC-oriented and didn’t use the Thinking Processes. Details were unavailable. Schragenheim reports that teams were often surprised by how broadly the revised understanding affected policy, but the paper supplies no outcome figures.
Consequences
Benefits. The pattern gives an organization a repeatable alternative to intuition, hindsight, and blame. It makes unexpected success learnable instead of treating only failure as worthy of investigation. The mixed team widens the explanation set, while explicit invalidation keeps the discussion from becoming a vote among stories. Connecting the revised assumption to policy makes the event part of POOGI, not an isolated retrospective.
The method also creates a durable organizational story. Sharing how the expectation failed helps other teams recognize the same assumption before another surprise forces the lesson.
Liabilities. The process can manufacture confidence if the team confuses a surviving hypothesis with a proven cause. It can also overcorrect by reversing an assumption that remains valid under ordinary conditions. The remedy in both cases is to preserve the boundary: state which prior cases supported the old assumption, identify what was different in this event, and record what evidence would change the conclusion.
The process isn’t a replacement for daily firefighting or chronic-problem analysis. If the cause and conflict are already recognizable, Daily Management with TOC may be the shorter route. If several persistent undesirable effects converge, use a CRT. Learning from Surprising Events is for the consequential exception that reveals the organization’s model was missing something.
Related Articles
Complements: Daily Management with TOC — Daily Management handles operating problems through an identifiable conflict; this process starts where a one-time surprise has no recognized cause or obvious conflict.
Supports: Fundamental Structure of ToC — The method operationalizes the structural emphasis on finding the flawed paradigm behind a limitation rather than merely treating the visible event.
Supports: Process of Ongoing Improvement (POOGI) — Learning from a consequential surprise feeds corrected assumptions and policies into the process of ongoing improvement.
Uses: Assumption — The process works backward from an unexpected outcome to the operating assumption that made the team's expectation reasonable but wrong in this case.
Uses: Categories of Legitimate Reservations (CLR) — Additional-cause and effect-cause-effect scrutiny help the team expand and invalidate candidate explanations without mistaking plausibility for proof.
Uses: Future Reality Tree (FRT) — Once the flawed assumption is understood, an FRT tests whether the proposed policy changes produce the intended future effects.
Sources
• Eli Schragenheim, “Systematic Learning from Significant Surprising Events: Drawing the Right Lessons for Improving the Future”, TOCICO-recognized Body of Knowledge paper (2022). Primary source for the eight-step method, team composition, illustrations, cautions about inference, implementation report, and policy-change goal.
• Eli Schragenheim and Avner Passal, “Learning from Experience: A Structured Methodology Based on TOC”, Human Systems Management 24(1), 95–104 (2005). Earlier peer-reviewed presentation of the method and its scope; cited here for existence and lineage because the full paper wasn’t available for this entry.
• Oded Cohen, “Daily Management with TOC,” in Cox and Schleier, Theory of Constraints Handbook (2010), Ch. 24. Source for the neighboring Cloud-based firefighting routine and the boundary drawn in this entry.