Upstream of Innovation

Opening

The title is the argument. The Imagination Machine — two words that, read with any care, refuse each other. A machine is, by definition, the opposite of imaginative. A machine follows rules. It produces outputs predictable from inputs. Imagination is the inverse: the capacity to produce something that no set of rules would have predicted, something genuinely new. The authors’ bet is that this contradiction is not rhetorical but architectural — that imagination can be embedded in a system that doesn’t crush it, and that the embedding is what separates organisations that renew from organisations that optimise themselves into irrelevance.

The Argument

The Imagination Machine (Harvard Business Review Press, 2021) rests on a specific structural claim about where imagination sits in the causal chain of organisational change. Most business books treat innovation as the upstream capability — the thing organisations need to cultivate. Reeves and Fuller place imagination before innovation in the causal chain. An organisation with strong innovation processes and weak imaginative input is optimising the wrong stage.

From this structural claim, the book derives a six-step cycle that describes how imagination actually moves through an organisation from initial stimulus to institutionalised reality: The Seduction, The Idea, Collision, Epidemic, The New Ordinary, and The Encore.

“Imagination is one of the least understood but most crucial ingredients of success.” — Martin Reeves & Jack Fuller

The Authors’ Vantage

Martin Reeves is Chairman of the BCG Henderson Institute, Boston Consulting Group’s internal think tank, and has been a senior strategy partner at BCG for over two decades. Jack Fuller is an expert in neuroscience and philosophy whose role is most visible in the book’s use of neuroscience as scaffolding for claims about how imagination works at the individual and organisational level.

Five Insights Worth Extracting

1. Imagination Is Upstream of Innovation

The book’s most useful move is to strip imagination of its romance. It is not inspiration descending from nowhere; it is the manipulation of mental models — the internal representations we carry of how the world works. Imagination begins when reality refuses to fit the model. That reframing matters because a mystical faculty cannot be managed, while a mechanical one can be triggered on purpose. The authors borrow the biologist’s notion of the Umwelt — the slice of reality an organism can perceive — to make the point that every firm operates inside a self-imposed perceptual cage, and that imagination is the act of rattling its bars.

2. Surprise Is the Raw Material — and It Arrives in Three Forms

If a broken model is the spark, surprise is the flint. Reeves and Fuller catalogue three doorways through which surprise enters: accidents (unintended consequences — Play-Doh began as a failed wallpaper cleaner), anomalies (the datum that will not behave, the “That’s funny…” that Asimov prized over “Eureka”), and analogies (a resemblance imported from a distant domain, as when Charles Merrill reimagined retail brokerage on the model of the self-service supermarket). The practical payload is that surprise can be hunted rather than awaited — a firm can deliberately expose itself to the accidental, the anomalous, and the analogical.

3. Ideas Must Collide with the World to Evolve

An imagined idea is a hypothesis, not a conclusion, and the book is firm that thinking alone will not refine it. The Collision (ACT) step insists on a fast loop between mind and world: build a rough version, let it collide with reality, and harvest the new surprises the collision throws off. This is where imagination stops being a private event and becomes a learning system. The authors are sympathetic to play, prototype, and provocation precisely because each generates the feedback that a purely analytical culture starves itself of. The loop, not the lightning bolt, is the engine.

4. The Hard Part Is Collective, and Then It Is Permanent

The EPIDEMIC (AMPLIFY) and THE NEW ORDINARY (CODIFY) steps confront the genuinely difficult work of turning one person’s idea into a shared mental model and then into an “evolvable script” — a corporate routine durable enough to capture value yet loose enough to be re-imagined. The deepest insight arrives last: imagination is not banked. SUSTAIN demands mental ambidexterity — running today’s model at scale while re-imagining tomorrow’s — which most firms treat as a sequence and the authors insist must be simultaneous.

5. The Six-Step Cycle Is Prescriptive About Order, Not Just Content

The step list looks on first read like a taxonomy. The sharper reading treats it as a sequence claim: imagination fails when any single step is skipped or compressed, and the specific failure modes correspond to the specific steps. The cycle’s utility is not as a checklist but as a diagnostic lens — when imagination fails in a specific organisation, the failure can usually be located at a specific step, and the remediation depends on which step broke.

The Imagination Cycle, Made Visible

Click on any numbered step to explore its Trigger, Attributes, Failure Mode, how to Rectify, and what Hands Over to the next step.

Where the Argument Echoes

In evolutionary biology, the trigger–rethink–act loop is recognisable as variation under selection. A surprise generates variants (imagined models); reality acts as the selection pressure that culls them. The parallel reveals what the book’s tidy six-step diagram obscures: most imagined ideas are supposed to die, and a healthy imagination machine needs a high mutation rate and a ruthless selection mechanism working in tandem.

In control engineering, the ACT step is a feedback loop, and feedback loops have a property the book under-stresses: latency. The longer the delay between acting and learning, the more the system oscillates or stalls. The parallel tells the reader that the rate-limiting variable in an imagination machine is not idea quality but loop speed.

In thermodynamics, the SUSTAIN problem is an entropy problem. Left alone, an organisation’s scripts ossify. The parallel reframes “sustaining imagination” not as motivation but as continuous energy input against decay — imagination is a pump that must keep running, not a reservoir to be filled.

Disciplines for Building the Machine

The book’s operational core, distilled into three postures a leader can hold — each opened up for the reader who wants more, and compressed for the reader who wants to remember.

Where the Argument Strains

Reeves and Fuller’s framework invites engagement from the disciplines it draws on rather than dismissal from outside them. Four lenses bring the argument’s genuine claims into sharper relief.

Through the lens of behavioural science, the book asserts that surprise reliably triggers imaginative rethinking — but the better-documented human response to a model-violating surprise is rationalisation, not revision. Festinger’s work on cognitive dissonance and the broad confirmation-bias literature suggest the default is to defend the old model, not break it. The framework needs, and largely lacks, an account of why the trigger step so often fails to fire.

Through the lens of organisational psychological safety, Amy Edmondson’s empirical work supplies the foundation Reeves and Fuller gesture at but do not fully build. The book’s claim that social architecture matters more than individual brilliance becomes specifically testable through Edmondson: Seduction dies when surprise cannot be reported without career risk; Collision dies when disagreement feels politically dangerous; Epidemic dies when championing a non-consensus idea is informally punished. The lens gives the framework an operational diagnostic the book does not quite construct for itself, and doing so makes the cycle more useful rather than less.

Through the lens of organisational theory, the central prescription — mental ambidexterity, running and reinventing at once — is real but famously hard, and the book underweights how hard. O’Reilly and Tushman document the structural separations, distinct funding, and protected leadership that ambidexterity demands. “Sustain imagination” is sound as an aspiration and thin as an instruction; the organisational plumbing is where most attempts actually break.

Through the lens of methodology, the cases are selected on the dependent variable. We meet the firms whose imaginative leap worked — Merrill, LEGO, Disney — and rarely the larger population that imagined boldly and failed. Without the denominator, the six-step process reads as a recipe when it may describe survivors. The reader should treat the cases as illustrations of a posture, not as evidence that the posture reliably pays.

Through the lens of AI and cognitive commoditisation, the book was written before the LLM explosion. Read now, the upstream-of-innovation claim has sharpened in ways the authors could not have anticipated. As AI commoditises the downstream stages of the cycle (pattern-matching Collision feedback, spreading through communication tools, institutionalising through documentation), the upstream stages — Seduction, The Idea, The Encore — become the only stages that remain distinctly human. The framework needs an extension the book did not write: which parts of Seduction and The Idea can AI perform, which require human judgement, and how the two integrate.

Through the lens of complex-adaptive-systems theory and emergence, the six-step cycle is, read through this lens, a description of emergence in organisational cognitive systems. Emergence theory has the formal apparatus to explain why the cycle must remain circular — why The Encore is not optional but structural. A system that reaches The New Ordinary and stops has lost its adaptive capacity. Reeves and Fuller discovered an emergence-theoretic principle and described it in consulting language; the emergence framing would defend the framework at a level of rigour the book itself does not reach.

Refracted through these lenses, the critiques above are not frictions to Reeves and Fuller’s perspective — they are scaffolding the reader can build around the book’s argument, finding places to anchor their own engagement and test the framework against disciplines it touches. What the authors built survives this engagement: imagination genuinely is upstream, genuinely is mechanical, and genuinely can be provoked. What the lenses expose are the framework’s honest edges — the trigger that resists firing, the ambidexterity that resists plumbing — which a careful reader treats as the work the book leaves them to finish.

Why This Book, Now

Three conditions make The Imagination Machine more consequential in 2026 than it was at publication in 2021. First, the AI industrialisation of analytical work has sharpened rather than weakened the book’s core claim. As deductive and inductive reasoning become commoditised — as AI systems become reliably competent at the tasks that most corporate analytical roles used to perform — the organisational premium on imagination has risen precisely because imagination is the part of cognitive work that AI systems remain genuinely bad at. The book’s 2021 claim that imagination would become the distinguishing organisational capability reads, in 2026, as one of the more prescient observations in the strategy literature of the early 2020s.

Second, the post-pandemic reset that many organisations underwent through 2020–2023 has made visible, with unusual clarity, how organisations actually behave when their settled patterns are disrupted — and the evidence is more mixed than the book’s crisis-favours-imagination claim suggests. Some organisations used the disruption as a genuine imaginative opening; most used it to fall back harder on the familiar. The book’s framework helps distinguish the two patterns.

Third, and most importantly for a reader on a thought-leadership site, the governance-of-imagination question has become newly urgent. As AI systems become capable of producing outputs that look imaginative — novel combinations, surprising proposals, unexpected syntheses — organisations face a genuine question about which parts of the imagination cycle remain distinctively human, and which are being outsourced to machines that can produce the form of imagination without its substance. The book, written before this question emerged in its current form, cannot address it directly. But the framework it provides is the clearest available tool for thinking through the question.

Closing

The durable question the book asks is whether organisations can be built to imagine systematically — whether the most-individual-seeming of human capacities can be made collective, cultivable, repeatable.

Which step of the Imagination Cycle is your organisation weakest at — and what would it cost, concretely, to strengthen it before the environment punishes you for the weakness?

Adjacent Reading