Three Horizons

Most planning jumps straight from “now” to “the goal” and skips the messy middle, or it freezes into a present-versus-future binary where one camp defends the status quo and another chases a disruption. Three Horizons refuses both. The durable cognitive move is to hold three overlapping change curves on one time canvas at once - a declining present system (H1), a contested transition zone (H2), and an emerging desired future (H3) - locate the actor inside that picture, and read the transition dynamics rather than treating the future as a single forecast or a single endpoint. Its signature, the part no neighboring method has, is reading the H2 middle in two directions: H2-plus innovations that genuinely carry the H3 future, and H2-minus innovations that get captured by H1 and merely prop up the incumbent system. The output is a three-horizons transition map: the named H1 system and what is failing in it, the H3 aspiration and the seeds of it already visible now, the H2 transition moves classified by which horizon they actually serve, and a read of where the actor is standing and where their attention and energy are currently going. The map is a shared orientation for dialogue, not a defended decision.

When to Use

A group is stuck in a present-versus-future binary - either defending the status quo or chasing a disruption - and needs to see that the managed decline of the old system and the cultivation of the new one are happening at once and have to be governed together.
The “messy middle” between today and the goal is being skipped, and the transition zone itself needs to be named, populated, and governed.
Incremental and transformative change are being funded or measured the same way, and they need to be distinguished so they are not held to the same metrics.
Actors with different mindsets - the H1 defender, the H2 entrepreneur, the H3 visionary - are talking past each other and need a shared map to talk across rather than past.
A faint “pocket of the future in the present” exists today and the question is how to grow it while the dominant system declines.

When NOT to Use

Do not use it to derive one backward path from a single fixed future. That is think-backcasting (fix one desired future, derive one linear milestone path back to a step you can take now). Three Horizons holds three curves at once and produces a present-state orientation map, not a route - and it adds the managed decline of H1 and the two-directional H2 read, which backcasting has no analogue for.
Do not use it to test which futures are plausible and whether a strategy is robust across them. That is think-scenario-planning (a set of divergent, equally-plausible external futures crossed from two critical uncertainties, held in parallel without ranking, with a robustness read). Three Horizons asserts one declining present and one desired future and reads the transition between them; it does not generate alternative external futures and must not be used to launder a single preferred narrative into a structured-looking picture.
Do not use it to trace the consequences of one change radiating outward. That is think-futures-wheel (one consequence map radiating from one change). Three Horizons runs along time with three curves, not outward from one event.
Do not use it to drive one problem down through layers of causation. That is think-iceberg-model (one problem driven down events, patterns, structures, mental models). Three Horizons is a horizontal time canvas, not a vertical depth probe.
Do not read the horizons as fixed time bands. Equating H3 with “5+ years out” and H1 with “now” is the single most common and most destructive failure. The horizons are degrees of transformation, not calendar distances: an H3 disruption can ship as fast as an H1 product. If the map is being built on a calendar axis, stop and rebuild it on degree-of-transformation.
Do not apply H1 governance to H2 and H3 work. Holding an early H3 venture to H1’s ROI and certainty hurdles strangles it. The entire value of the frame is that the three horizons need different metrics; using it without that discipline is worse than not using it.

Instructions

When asked to map a transition, govern an incumbent-to-future shift, or break a present-versus-future binary, follow these steps:

State the system in transition and the actor. Name, in one line, the system or domain whose transition is in question and the actor whose vantage point the map is drawn from. The map is always drawn from somewhere; say from where.
Set the axis to degree of transformation, not calendar time. Make explicit that the three curves are degrees of transformation (how far from today’s dominant logic), not time bands. This warning is load-bearing - if the reader places the horizons on a calendar, the tool collapses.
Describe H1, the declining present. Name the currently dominant system - the way things are done now - and, specifically, what is failing in it or losing fit as the environment changes. H1 starts high and declines; capture both that it dominates now and why it is under pressure.
Describe H3, the emerging desired future. Name the qualitatively different, desired future system, and identify the “pockets of the future in the present” - the faint seeds of H3 already visible today. H3 is an aspiration with present-day evidence, not a fantasy with no foothold.
Populate H2, the contested transition zone. List the disruptions, ventures, and innovations that bridge H1 and H3 - the moves actually in play or available in the messy middle. This is the object most planning skips; populate it explicitly.
Read H2 in two directions (the signature move). Classify each H2 move as H2-plus (genuinely carries the H3 future) or H2-minus (gets captured by H1 and merely props up the incumbent). A move can look innovative and still be H2-minus. This two-directional read is the heart of the method; do not skip it.
Locate the actor and their energy. Place the actor on the canvas: which horizon are they standing in, and where is their attention and energy actually going right now (defending H1, working the H2 middle, cultivating H3)? Name any mismatch between where they say they are and where their effort flows.
Note the governance implication. State, briefly, that the three horizons need different metrics, and flag where H1 governance is being (or would be) misapplied to H2 or H3 work. This is an orientation note, not a decision.
Emit the three-horizons transition map per references/TEMPLATE.md: the named H1 and what is failing, the H3 aspiration and its present seeds, the H2 moves classified H2-plus / H2-minus, and the actor located in the picture. Frame the map as a shared orientation for dialogue, never as a forecast or a defended answer. The template’s pre-printed evidence caveat is part of the artifact; carry it through verbatim.

Output Format

Use the template in references/TEMPLATE.md. The deliverable is the filled three-horizons transition map - the declining H1 and its failing fit, the emerging H3 and its present seeds, the H2 transition zone with each move classified H2-plus or H2-minus, and the actor located on the canvas with their energy read - not a prose essay. Keep the axis on degree of transformation, never on calendar time, and present the map as a sensemaking scaffold, not a decision.

Quality Checklist

Before finalizing, verify:

Evidence

Tier C (governing, conceptually plausible but undertested). Three Horizons is a real, named, twice-originated method - a growth-portfolio version (Baghai, Coley and White, The Alchemy of Growth, 1999) and a futures-practice version (Curry and Hodgson 2008; Sharpe et al. 2016) - with a clear lineage and a sustained academic and practitioner literature. That literature is genuine, and it is easy to mistake for effectiveness evidence: it is not. The peer-reviewed papers are method exposition and theory-building, and the best empirical entry (Luederitz et al. 2024, Ecology and Society) is a single descriptive case reporting that participants moved through transformative-learning phases in the predicted sequence - uncontrolled, single-case, with an outcome internal to the method’s own logic. There is no controlled or comparative study locatable measuring whether Three Horizons produces better foresight, better decisions, or better outcomes against any baseline, and a live external critique (Steve Blank, 2019) argues a core assumption - that the horizons are time bands - is wrong for modern innovation. So the evidence supports “well-described, twice-originated, plausibly useful as a scaffold,” and does NOT support any effectiveness claim. Transfer caveat: every piece of evidence is from human participants in workshops, strategy rooms, and field cases; none studies Three Horizons performed by or with an AI agent, so the evidence is transferred from human facilitation contexts and is not validated for agent-run use. The skill ships honestly as a C-tier sensemaking scaffold that names the transition zone and surfaces its dynamics, never as a predictor or an outcome method. Full grading, named sources, excluded figures, and the distinctness proof: evidence/dossier.md.

Examples

See references/EXAMPLE.md for a completed three-horizons transition map on a real decision.

Deep dive: worked example

A full worked run (the shared Northwind scenario)

Three-Horizons Transition Map - Worked Example

A completed run of the three-horizons skill on a real, consequential decision. This is the quality bar a generated transition map should meet.

Uses the shared recurring scenario (Northwind, a B2B SaaS weighing a self-serve free-tier launch). Where think-scenario-planning builds several uncontrollable external futures and asks which free-tier moves survive all of them, and think-backcasting fixes one desired future and maps the path back, this skill does neither: it holds Northwind’s declining sales-led present, a contested transition middle, and an emerging product-led-plus-AI-native future as three curves at once, and reads whether each transition move carries the future or merely props up the incumbent. See docs/internal/AUTHORING.md.

The axis below is degree of transformation, not calendar time. The map is a shared orientation for the team’s dialogue, not a forecast and not a decision. The three horizons need different metrics.

System in transition and actor

System / domain in transition: how Northwind acquires, converts, and retains customers - the shift from a sales-led, seat-licensed software business toward a product-led, increasingly AI-native one.
Actor (vantage point): Northwind’s leadership team, deciding whether and how to launch a self-serve free tier.
Axis reminder: H1, H2, and H3 below are degrees of transformation away from Northwind’s current sales-led logic, NOT time bands. The H3 future could arrive fast or slow; an AI-native competitor could collapse the workflow overnight.

H1 - the declining present (dominant now, losing fit)

The dominant system today: Northwind sells top-down. Account executives run demos, procurement negotiates annual seat licenses, and revenue is booked as multi-seat contracts. This is how the company makes its money now, and it works.
What is failing in it / losing fit: Buyers increasingly expect to try before they buy; bottom-up adoption is bypassing the AE motion in Northwind’s category; sales-led CAC is rising as deals get more scrutinized; and the seat-license logic fits poorly with usage-based, AI-assisted workflows that do not map cleanly to “a person at a desk.” The core still pays the bills but is losing fit with how software is now bought and used.

H3 - the emerging desired future (faint now, growing)

The desired future system: a product-led, AI-native Northwind where the product itself acquires and expands accounts - users self-serve in, value lands before any human conversation, and the workflow is increasingly done by the product (agents doing the job) rather than in it (a UI a person operates). Monetization follows usage and outcomes, not seats. This is a different logic, not a faster version of the sales motion.
Pockets of the future in the present: a handful of inbound users already sign up, configure themselves, and reach value with zero sales touch; the most-loved part of the product is already an automation that does work for the user; a small cohort expands usage on its own and only later talks to sales. These faint signals are H3 already alive inside H1.

H2 - the contested transition zone (the messy middle, read in two directions)

The moves available to bridge the sales-led present and the product-led, AI-native future - classified by which horizon each actually serves. A move can look like progress and still be H2-minus.

H2 move (disruption / venture / innovation)	Classification	Why it serves that horizon
A self-serve free tier built as a true product-led on-ramp (real time-to-value with no human, instrumented activation, self-serve upgrade path)	H2-plus (carries H3)	Builds the product-led acquisition and expansion muscle the H3 future needs; the free tier is the wedge that grows the pocket of the future.
The “same” free tier built as a lead-gen funnel for the AE team (signups become SQLs, no real self-serve value, sales still closes everything)	H2-minus (captured by H1)	Looks like product-led growth but only feeds the existing sales machine; it props up H1 and builds no new capability, while consuming the brand and roadmap budget a real PLG bet would need.
Rebuild the core workflow to be AI-native (agentic, outcome-priced)	H2-plus (carries H3)	Directly grows the “done by the product” seed of H3 and hedges the step-change risk that an AI-native entrant collapses the workflow.
Bolt AI features onto the existing seat-licensed UI as upsell line-items	H2-minus (captured by H1)	Monetizes AI inside the seat-license logic; reinforces H1’s pricing and packaging rather than moving toward usage/outcome-based H3.
Make the product modular and embeddable (be where buyers already are)	H2-plus (carries H3)	Keeps the path to a product-led, agent-surfaced future open regardless of how buying organizes.

H2-plus moves (genuinely carry the H3 future): a free tier built as a real product-led on-ramp; an AI-native rebuild of the core workflow; a modular, embeddable architecture.
H2-minus moves (captured by H1, prop up the incumbent): a free tier run as AE lead-gen with no self-serve value; AI features sold only as seat-license upsells. Both feel like motion and quietly entrench the declining system.

The actor located on the canvas

Where the actor is standing: Northwind’s leadership is standing firmly in H1 - the sales-led system is where the revenue, the headcount, and the quarterly targets live - while debating an H2 move (the free tier) and gesturing at an H3 future.
Where attention and energy actually go right now: almost all energy goes to defending and hitting the H1 number (AE pipeline, this quarter’s bookings). The free-tier conversation is real but under-resourced, and the AI-native rebuild is “next year.” The mismatch to name: the team says it wants the product-led future, but its attention, comp plans, and roadmap weight all flow to H1 - which is exactly the gravitational field that turns an H2-plus free tier into an H2-minus lead-gen funnel.

Governance implication (orientation note, not a decision)

The three horizons need different metrics, and the most important sensemaking point on this map is that Northwind is at risk of governing its H2 free-tier bet with H1 rules. If the free tier is measured by SQLs handed to AEs and by this quarter’s bookings (H1 metrics), it will be steered toward H2-minus regardless of the intent stated in the strategy deck - activation, self-serve conversion, and product-qualified expansion are the H2-plus metrics, and they will look like failures against an H1 scorecard for a while. The dialogue this map should open is not “should we launch a free tier” but “are we prepared to govern it - and the AI-native rebuild - on different metrics than the core, and to protect that energy from the H1 gravity well.” That is an orientation for the team’s conversation, not a recommendation this map makes for them.

Note how this differs from its neighbors on the same Northwind decision. The think-scenario-planning example constructs four alternative external worlds Northwind does not control and asks which free-tier moves survive all of them (parallel futures, robustness). The think-backcasting example fixes one desired future and maps the milestone path back to today (one route). This three-horizons map does neither: it holds the declining sales-led present, the contested transition middle, and the emerging product-led future as three simultaneous curves, classifies each transition move as carrying the future or propping up the incumbent, and locates where the team’s energy actually flows. The deliverable is a shared orientation on the transition and its H2-plus / H2-minus dynamics, not a set of worlds and not a path.

Grounding: the full evidence dossier

What the research does and does not show, with graded sources

Evidence Dossier: Three Horizons

The single source of truth for the three-horizons skill. The SKILL.md, the sidecar (skill.meta.yml), and the eval cases all derive from this file. If a claim is not here, it does not belong in the skill. Promoted from frameworks/_proposed/three-horizons/dossier.md and admitted as a Build at tier C (honoring the catalog’s prior cand / build / C tag - not inflated).


Skill	`thinking-framework-skills.three-horizons` (installable name `think-three-horizons`)
Family	systems-and-consequences
Evidence tier	C governing (conceptually plausible but undertested - see “What the evidence shows”)
Confidence	Low-to-moderate that the three-curve transition map breaks present-versus-future binaries and surfaces the H2 transition dynamics as a sensemaking scaffold; no confidence in any decision-outcome or foresight-accuracy effect, for which no controlled evidence exists
Status	draft (admitted as a weak Build at tier C; the maintainer could legitimately keep this as `cand` rather than promote)

1. The mechanism (what actually does the work)

Three Horizons lays three overlapping change curves on a single time canvas and asks where the actor is sitting between them.

Horizon 1 (H1) is the currently dominant system - the way things are done now - which starts high and declines over time as it loses fit with a changing environment.
Horizon 3 (H3) is a qualitatively different, desired future system that starts as a faint pocket of the present (a “pocket of the future in the present”) and grows as the environment shifts toward it.
Horizon 2 (H2) is the contested transition zone between them - the disruptions, ventures, and innovations that bridge the gap - read explicitly in two directions: H2-plus innovations that genuinely carry the H3 future, and H2-minus innovations that get captured by H1 and merely prop up the incumbent system.

The durable cognitive move is: hold three overlapping change curves - a declining present system, a contested transition zone read in two directions, and an emerging desired future - on one time canvas, locate the actor within that picture, and read the transition dynamics between them. Two things distinguish it from ordinary planning. First, it refuses the present-versus-future binary: the managed decline of H1 and the cultivation of H3 are governed as one picture, not as a fight between camps. Second, the H2-plus / H2-minus split is its signature object - the messy middle most planning skips is named, populated, and classified by which horizon each move actually serves.

The emitted artifact is the three-horizons transition map: the named H1 system and what is failing in it, the H3 aspiration and the seeds of it visible now, the H2 transition moves classified by which horizon they actually serve, and a read of where the actor is standing and where their attention and energy are currently going. The output is a shared orientation for dialogue and “future consciousness,” not a defended decision - its own authors frame it as a structure for sensemaking, not a method that yields an answer.

2. Lineage

The honest account separates two distinct lineages that share the name and the three-curve picture but were built for different jobs.

The McKinsey growth-portfolio version - owed to Mehrdad Baghai, Stephen Coley and David White, McKinsey consultants, in The Alchemy of Growth: Practical Insights for Building the Enduring Enterprise (Perseus, 1999); McKinsey’s own “Enduring Ideas: The three horizons of growth” is the canonical short statement. Here the three horizons are an investment-portfolio frame: H1 = defend and extend the profitable core, H2 = build emerging high-growth businesses, H3 = seed options for the future. The job is balancing a portfolio of investments across maturity stages so the pipeline does not run dry while the core is harvested. This is the version most managers mean, and the version Steve Blank attacked.
The IFF / futures-practice version - owed to the International Futures Forum circle. Read Andrew Curry and Anthony Hodgson, “Seeing in Multiple Horizons: Connecting Futures to Strategy,” Journal of Futures Studies 13(1) (2008): 1-20, the foundational method paper; Bill Sharpe, Anthony Hodgson, Graham Leicester, Andrew Lyon and Ioan Fazey, “Three Horizons: A Pathways Practice for Transformation,” Ecology and Society 21(2) (2016): art. 47, for the transformation-pathways framing; and Sharpe’s book Three Horizons: The Patterning of Hope (Triarchy Press, 2013) for the practitioner treatment. Here the curves are patterns of human intention and systemic change; the point is to develop “future consciousness” and to make the H1-to-H3 transition and its power dynamics explicit and dialoguable. The job is sensemaking and shared orientation under deep uncertainty, not portfolio allocation.

Both reduce to the same durable move, which is what this skill implements. “Three Horizons” is a generic descriptive term in common use across both lineages, with no registered trademark and no single owner - it is attributed (Baghai/Coley/White for the growth version; Sharpe and the IFF circle for the foresight version), not branded.

3. What the evidence shows, and what it does NOT show

The honest grade is C (conceptually plausible but undertested), and this dossier honors the catalog’s prior C tag rather than overturning it. The reasoning has to be careful, because Three Horizons has genuine peer-reviewed literature - which is easy to mistake for effectiveness evidence, and is not.

What the record supports. Three Horizons is a real, named, twice-originated method with a clear lineage and a sustained academic and practitioner literature. The futures-practice version appears in peer-reviewed venues: Curry and Hodgson (2008) in the Journal of Futures Studies set out the method and its integration with other futures tools; Sharpe, Hodgson, Leicester, Lyon and Fazey (2016) in Ecology and Society presented it as a transformation-pathways practice. There is at least one published descriptive case study - Luederitz and colleagues (2024) in Ecology and Society, on climate action with SMEs in a Canadian Biosphere Reserve - reporting via interviews and workshops that participants moved through transformative-learning phases in the predicted sequence. That is the high-water mark of the evidence, and it is the ceiling on the grade.

What the record does NOT support. None of this is effectiveness evidence in the sense this library grades on. There is no controlled or comparative study locatable - no randomized or quasi-experimental test, no comparison against another method or a no-method baseline - measuring whether using Three Horizons produces better foresight, better decisions, or better outcomes. The peer-reviewed papers are method exposition (2008, 2016) and theory-building; the 2024 study is a single descriptive case with no control, no counterfactual, and an outcome (self-reported perspective change) that is internal to the method’s own logic rather than an independent result. The framework’s own most-cited co-author, Andrew Curry, wrote a paper (“Searching for systems: understanding Three Horizons”) partly responding to critics who “couldn’t see what was distinctive about the model,” which tells you the distinctiveness was contested even inside the field. And the most prominent external critique, Steve Blank’s, argues the model’s central temporal assumption is wrong for modern innovation. A method that is well-described, twice-originated, and widely taught, but has zero controlled validation and a live dispute about both its distinctiveness and a core assumption, is the definition of tier C: conceptually plausible, plausibly useful as a scaffold, and undertested - not tier P, because even the practitioner-effectiveness base (the reported-application track record that lifts, say, theory-of-constraints to P) is thin and case-anecdotal here rather than broad.

Excluded-figures note. No specific effect size, success rate, or adoption statistic is asserted as fact in this dossier, because none traces to a nameable primary source measuring the method. The widely-repeated practitioner claims that Three Horizons “improves strategic thinking” or “drives transformation” appear without any controlled measurement behind them and are excluded from the grade under the evidence rule; they are marketing and testimony, not findings.

4. Transferred-evidence flag (required honesty for this library)

Every piece of evidence above is from human participants in workshops, strategy rooms, and field cases. None of it studies Three Horizons performed by or with an AI agent. The evidence is transferred from human facilitation contexts and is not validated for AI-augmented or agent-run use. The AI value is mechanical and modest: an agent makes the method cheap to run, forces the discipline (set the axis on degree of transformation rather than calendar time, name what is failing in H1, find genuine present-day seeds of H3, populate the H2 middle, and classify each H2 move H2-plus or H2-minus), and produces a durable, inspectable artifact - benefits that do not depend on any contested outcome claim. The skill ships honestly as a C-tier sensemaking scaffold with a hard “this is an orientation, not a forecast or a decision” wall, never as a predictor.

5. When it works / when it fails (drives the eval negative cases and “When NOT to Use”)

Works best when:

A group is stuck in a present-versus-future binary - either defending the status quo or chasing a disruption - and needs to see that the managed decline of the old system and the cultivation of the new one are happening at once and must be governed together.
The transition zone needs to be named explicitly (most planning tools jump from “now” to “the goal” and skip the messy middle).
Incremental and transformative change must be distinguished so they are not funded or measured the same way.
Actors with different mindsets (the H1 defender, the H2 entrepreneur, the H3 visionary) need a shared map to talk across rather than past each other.

Fails or misleads when (poor-fit / anti-patterns):

The horizons are read as fixed time bands. The single most common failure, and the core of Steve Blank’s critique: teams equate H3 with “5+ years out” and H1 with “now,” then misclassify and misgovern. In reality a Horizon 3 disruption can ship as fast as a Horizon 1 product (Uber repurposed existing technology into a new model overnight); the horizons are degrees of transformation, not calendar distances, and the time-band reading destroys the tool.
H1 governance is applied to H2 and H3 work. Holding an early H3 venture to H1’s ROI and certainty hurdles strangles it. The frame’s value is precisely that the three horizons need different metrics; using it without that discipline is worse than not using it.
The map substitutes for evidence about which future is likely. Three Horizons asserts one desired H3 and one declining H1; it does not test whether that H3 is the future that will actually arrive, nor does it generate alternative external futures. Used alone it can launder a single preferred narrative into a structured-looking picture. When the question is “which futures are plausible and is my strategy robust across them,” it is the wrong tool (that is scenario work).
It is treated as analysis rather than as a sensemaking scaffold. The output is a shared orientation, not a decision. Its own authors frame it as a structure for dialogue and “future consciousness,” not a method that yields a defended answer.

6. Distinctness (why this is a Build, and a weak one)

Verdict: Build, at the honest floor of tier C. The Build burden is to name one distinct, durable cognitive move that no shipped skill produces and that is not a clean chain of shipped skills - and Three Horizons clears it, narrowly, on the artifact, while remaining a weak-evidence entry the maintainer could legitimately keep as cand.

The distinct move is: hold three overlapping change curves - a declining present system, a contested transition zone read in two directions, and an emerging desired future - on one time canvas, locate the actor within that picture, and read the transition dynamics between them. The emitted artifact is the three-horizons transition map. The closest shipped skills, and why none subsumes that move:

think-backcasting is the nearest neighbor, and the one that most threatens the Build. Both anchor on a desired future. But backcasting fixes a single H3 and derives one linear milestone path back from it to the next step now; its artifact is a path. Three Horizons does not derive a path - it holds three curves at once, explicitly maps and classifies the H2 transition zone (the H2-plus / H2-minus split is its signature, and has no analogue in backcasting), reads the managed decline of H1 as a parallel concern, and produces a present-state orientation map rather than a route. The shared element (a vivid desired future) is real but is well under a fifth of the working mechanism; backcasting has no transition-zone object and no three-curve simultaneity. Not a fold.
think-scenario-planning is the second-nearest, and it fails the overlap test on the object. Scenario planning constructs a set of divergent, equally-plausible external futures (the 2x2 critical-uncertainty matrix) and robustness-tests strategy across them; its artifact is a set of parallel worlds held without ranking. Three Horizons produces the opposite shape: one structured transition picture with a single declining present and a single desired future, focused on the dynamics of getting from one to the other. One is “which futures might arrive, and is my strategy robust” (parallel, uncommitted); the other is “how does the present give way to the future I want, and what is in the contested middle” (single transition, committed direction). Different object, different artifact.
think-futures-wheel and think-iceberg-model are low-overlap by geometry. Futures-wheel radiates consequences outward from one change (radial, consequence-mapping). Iceberg-model drives one problem down four levels of causation (vertical, depth). Three Horizons runs along time with three curves (temporal, transition-mapping). Neither produces the three-horizons artifact.

It is also not a recipe: you cannot assemble the three-curve transition map by running backcasting then futures-wheel then something else; the H1-decline / H2-contested-transition / H3-emergence canvas, with the actor located in it and the two-directional H2 read, is a single integrated apparatus, not an emergent property of a fixed chain. So Build, not Recipe; and no shipped mode produces it, so Build, not Fold.

The honest caveats that keep this a weak Build. First, the evidence is C, the floor of what the library will ship, weaker than the P of most shipped skills and weaker than scenario-planning’s and backcasting’s P. Second, the distinctness, while real, rests substantially on the H2 transition-zone object and the three-curve simultaneity; strip those and what remains (a desired future plus a declining present) drifts toward backcasting. The skill leads with the transition-zone classification and the time-is-not-the-axis warning precisely so it does not collapse into a fuzzier backcast.

7. Named sources

Mehrdad Baghai, Stephen Coley and David White, The Alchemy of Growth: Practical Insights for Building the Enduring Enterprise (Perseus Publishing, 1999). The origin of the growth-portfolio three horizons (defend core / build emerging / seed options). Foundational practitioner book; no controlled evidence, a synthesis of consulting case experience. (P, for the practitioner origin - not effectiveness evidence)
Andrew Curry and Anthony Hodgson, “Seeing in Multiple Horizons: Connecting Futures to Strategy,” Journal of Futures Studies 13(1) (2008): 1-20. The foundational futures-practice statement of the method; exposition and integration with other futures tools, explicitly noting the method is “still in development.” Peer-reviewed method paper, not an effectiveness study. (C)
Bill Sharpe, Anthony Hodgson, Graham Leicester, Andrew Lyon and Ioan Fazey, “Three Horizons: A Pathways Practice for Transformation,” Ecology and Society 21(2) (2016): art. 47. Presents Three Horizons as a transformation-pathways practice for working with complexity and developing “future consciousness.” Peer-reviewed practice/theory paper; descriptive, no controlled comparison. (C)
Christopher Luederitz et al., “Theorizing how the Three Horizons approach supports transformative learning: insights from advancing climate action in a Canadian Biosphere Reserve,” Ecology and Society 29(2) (2024): art. 18. A single descriptive case (interviews + workshops with SMEs) reporting participants moved through transformative-learning phases in the predicted order. The best available empirical entry, but uncontrolled, single-case, with a within-method outcome - it does not measure effectiveness against any baseline. (C)
Steve Blank, “The Fatal Flaw of the Three Horizons Model” (steveblank.com, 2019; later retitled “Fast Time in Three Horizon High” after pushback from an original author). Argues the horizons are not time-bound and that H3 disruptions can ship as fast as H1 products, so the temporal reading misleads. The most prominent external critique; opinion/practitioner essay, cited to mark the live dispute over a core assumption. (A, as a critique)
Andrew Curry, “Searching for systems: understanding Three Horizons.” An in-field reckoning written partly in response to critics who “couldn’t see what was distinctive about the model”; useful as primary evidence that the method’s distinctiveness was contested within futures studies. Practitioner/scholarly essay. (P, as lineage and critique - not effectiveness evidence)

Excluded on the evidence rule: no effect size, success rate, or adoption statistic is asserted as fact, because none traces to a nameable primary source measuring the method. Practitioner claims that Three Horizons “improves strategic thinking” or “drives transformation” are testimony and marketing, not findings, and do not enter the grade.

Was this page helpful?

Thinking Framework Skills v0.8.0 · 56 frameworks