Speculative Design for Innovation Teams: Filling Horizon 3 and Killing Ideas Cheaply

Speculative design gives innovation teams the raw material Horizon 3 is supposed to contain: futures ten or more years out, built from real signals and made specific enough to generate ventures from. It works equally well in reverse, as a cheap way to kill ideas. A concept set in 2040 and backcast to the present shows which ideas in the funnel depend on a world nobody is building, before anyone pays for a pilot.

Innovation teams have the mandate, the budget line and the portfolio slot marked "long term". What they rarely have is a way to fill that slot with anything other than the core business's next idea, presented with more confidence. For the definition, start with what speculative design is; this guide is about where it sits in an innovation function's machinery.

What innovation teams already have

More than most. Innovation teams run idea funnels and intake forms, hackathons and design sprints, venture studios with investment committees, business model canvases, and some version of the stage-gate: the sequence of development stages separated by decision points that Robert Cooper set out in Business Horizons in 1990. Many organise their portfolio around the three horizons model, introduced by McKinsey's Mehrdad Baghai, Stephen Coley and David White in The Alchemy of Growth (1999). Horizon 1 extends and defends the core business, Horizon 2 builds emerging ones, and Horizon 3 seeds options for growth further out.

That is a head start. The team already accepts that a portfolio holds bets with different time frames and different standards of evidence. It prototypes quickly and is not frightened of showing something unfinished. And it already has gates, formal moments where an idea is judged, which is exactly where the output of speculative work needs to arrive. The workflow we teach is deliberately shaped like the processes innovation teams already run; the full methodology explains why.

Why Horizon 3 keeps emptying

Horizon 3 is where an innovation team is meant to work, and the organisation keeps pulling it out.

Part of the problem is the horizon itself. When organisations attached dates to the model, Horizon 3 meant something like three to six years. Steve Blank, writing in 2019, gives 36 to 72 months as a typical example, and argues that disruptive ideas now arrive as fast as incremental ones. By the standards of speculative design, six years is still inside the planning window. Today's platforms, contracts and competitors are all still binding, so the long-term ideas turn out to be adjacencies.

The larger part is governance. Funnel gates ask for evidence of demand, a market size and a business-unit sponsor. Horizon 1 ideas can supply all three; Horizon 3 ideas cannot, by definition. Over a few review cycles the slot fills with Horizon 2 ideas in Horizon 3 clothes, and the team spends its year running pilots for the core. Hackathons do not fix this: forty-eight hours is enough to prototype an improvement to the present and far too short to build a world for anything else. Meanwhile the range of what could happen within a decade keeps widening, which we call possibility dilation, so the ground a Horizon 3 scan must cover grows while the slot shrinks.

Speculative design helps on both fronts. It produces Horizon 3 material that does not depend on present-day demand evidence. And it offers a different test for Horizon 3 ideas — not how big the market is, but which changes in the world the idea depends on, and whether anyone is making them.

Where it fits in the innovation pipeline

A worked example: an outdoor brand after "forever chemicals"

Imagine an innovation team at a mid-sized European outdoor clothing brand; call it Calder. Its Horizon 3 slot holds three ventures: a trail-planning app, a branded resale platform and a "storm-proof for life" guarantee on its best shell jacket. On inspection, the core business could launch any of them next year. The board has asked for a real one.

The brief. Source of change: the restriction of PFAS, the "forever chemicals" long used for water and stain repellency, as a class. Domain: weather protection in outdoor clothing. Horizon: 2040, chosen to sit well past the furthest regulatory date anyone at Calder currently plans around. Audience: the venture board.

Signals. On 20 December 2022 3M announced that it would exit PFAS manufacturing and work to discontinue PFAS across its product portfolio by the end of 2025. On 7 February 2023 the European Chemicals Agency published a proposal from the national authorities of Denmark, Germany, the Netherlands, Norway and Sweden to restrict PFAS under REACH as a group of around 10,000 substances. And France's law of 27 February 2025 banned the manufacture, import and sale of clothing and footwear containing PFAS from 1 January 2026, protective gear excepted, widening to all textiles from 2030. A supplier leaving, a regulator proposing, a legislature acting: each a who, a when and a what.

Future wheel. At the centre, the restriction complete: in 2040 no clothing sold anywhere Calder trades contains PFAS. The first ring, reformulated finishes and their cost, is already in every supplier's compliance plan. The second ring is where it becomes an innovation question. Water repellency stops being a permanent property of a jacket and becomes something maintained, so performance depends on what happens after the sale and the brand's relationship with the garment continues past the till. Stockists start to look like workshops. In the third ring, re-proofing becomes an autumn ritual in the way bike servicing is, and Calder's real competitors turn out to be independent repairers.

The concept. Calder Season Stamp. Tagline: "Waterproof is something you keep up." Every Calder shell has a blank woven label inside the collar. Each autumn the owner drops the jacket at any stockist; it comes back within a week cleaned, re-treated with a PFAS-free finish, tested on a spray rig, and stamped with the date and the rating it achieved. Sketch: the inside of a collar, a label with five rows of dated stamps.

Implications. For the owner, buying a jacket becomes the start of an annual appointment. Socially, people who skip the stamp wear jackets that quietly fail, so being dry on a hill becomes partly a matter of income. Culturally, a label full of stamps becomes a status marker for long ownership, reversing the prestige of the new. Ethically, old PFAS-treated jackets come back over the counter, and their disposal becomes Calder's problem. Economically, revenue moves from new jackets to service, and stockist margins and floor space are renegotiated. How it goes wrong: a freshly stamped jacket fails in a storm, and the label becomes the evidence in the claim.

Backcasting. Before 2040: the EU restriction proposed in 2023 adopted with clothing in scope. A spray test that stockists can run in store and customers trust. A PFAS-free finish that lasts a full season and can be applied by a trained stockist. Shops willing to take garments in, the way bike shops take bikes. And one rung in Calder's own hands: take-back logistics. Most of these are institutional, and institutional rungs are where an innovation team can actually open a conversation. Our essay on backcasting covers the method.

What changes this quarter. The Season Stamp does not enter the funnel; it is not a venture yet. Its job is to test the ventures that are there. The "storm-proof for life" guarantee was costed on the assumption that a jacket's protection is built in at the factory and lasts. In the backcast world protection is maintained, so the guarantee is an open-ended liability. It is killed in a week, at the cost of a workshop rather than a pilot. A small Horizon 3 bet enters instead: one season of re-proofing at two stockists, recording what customers bring back and how the jackets test. The team opens a conversation with a textile testing lab about an in-store test. And the stage-gate gains a permanent question for every Horizon 3 submission: which change in the world does this depend on, and who is making it?

Where it goes wrong for innovation teams

Innovation theatre with better props. The futures jam becomes the new hackathon: an offsite, a wall of renders, a post about it, and nothing entering the funnel. Theatre is judged by visibility. This work is judged by the decisions it changes, and if none are named before it starts, theatre is what it will be.

Putting the concept through the gate. A 2040 concept submitted to a stage-gate that asks for market sizing will either die for the wrong reason or teach the team to invent numbers. The concept's job is to generate and kill ventures, not to be one.

No one to absorb it. Innovation teams sit outside the line, so their output depends on a business unit choosing to pick it up, usually one whose roadmap is already locked. This is how good work becomes orphan insight. The remedy from our seven failure modes applies directly: name the receiving function, and the point in its cycle where the work lands, before the work begins.

Counting the funnel. Ideas generated and pilots launched reward volume, and speculative work done well often makes the funnel smaller. Measure it as we describe in measuring speculative design ROI: question quality, decision shifts, optionality, and avoided cost, which is where a documented kill belongs.

A first exercise for your next portfolio review

1. 1.List every idea in your Horizon 3 slot. Next to each, write the year in which it would make most of its money. Anything inside five years moves to Horizon 2 on paper.
2. 2.Choose one source of change your sector already talks about. Find three signals for it, each a named, dated event rather than a trend headline.
3. 3.Run a two-hour future wheel with the team, set in a fixed year at least ten years out and written at the top of the board.
4. 4.Generate concepts silently and alone for fifteen minutes before anyone speaks. Choose one and give it a name, a tagline, a description and a sketch.
5. 5.Backcast it: list the institutional and regulatory changes that had to happen first.
6. 6.Hold that list against every idea in your funnel. Any idea that depends on one of those changes not happening is a candidate to kill or reshape.
7. 7.Record one decision (a kill, a reshape, a partnership conversation, a new gate question) with the date, in a log you will revisit in six months.

Where to learn it as an innovation team

Speculative Design Basics is a self-paced online course of 42 lessons with lifetime access, so it fits around a team's delivery commitments. For innovation teams the Tools & Methods chapters do most of the work, covering signals, future wheels, future prototyping and backcasting among others. The Creating Value chapter covers how the practice relates to foresight and strategy, product and service development, and leadership: the three functions an innovation team sits between. The case studies, including work by Superflux, Extrapolation Factory and Near Future Laboratory, show what the output looks like at professional fidelity. Work through the first ten lessons, and if they have not improved your understanding and potential for impact, we refund in full. Teams that sit close to product or planning should also read our guides for product managers and strategists; the full series is at speculative design for every role.

Key takeaways

• •Speculative design gives innovation teams genuine Horizon 3 material: futures ten or more years out, built from dated signals and made into specific concepts.
• •Used in reverse, it kills ideas cheaply by showing which funnel ideas depend on changes nobody is making.
• •Horizon 3 empties because funnel gates demand evidence only near-term ideas can provide; the fix is a different gate question, not a braver team.
• •A speculative concept is not a venture, and putting it through a stage-gate destroys it.
• •Name the receiving function before the work begins, or the output becomes orphan insight.
• •Measure decisions changed, not ideas generated; a documented kill is an avoided cost.