Speculative design gives architects and urban designers a way to test a building against the decades it will stand through, rather than the day it opens. In place of one sunny render of handover day, the method produces specific artefacts from the building's middle age — a render of its worst ordinary day, a notice pinned up in its lobby — and works back from them to the decisions that are still cheap to change at Stage 2.
Architecture is already a speculative profession. Every planning application argues for a place that does not exist yet, and every render is a picture of the future. But the architectural render is speculation with the criticism taken out. It is always mid-afternoon, the trees are mature on day one, nobody is too hot, and the ground-floor unit is a café with its lights on. It shows the future the scheme is selling, which is the one future guaranteed not to arrive.
This guide covers only what is particular to applying speculative design to buildings and places. The general sequence, from brief to backcast, is in our practitioner's workflow.
What architects already bring to speculative design
More than most disciplines that come to this work. Architects already think in time. Frank Duffy's idea of shearing layers, which Stewart Brand expanded in How Buildings Learn (1994) into site, structure, skin, services, space plan and stuff, treats a building as parts that change at different speeds. "Long life, loose fit, low energy", the phrase Alex Gordon, a former RIBA President, gave the profession in the 1970s, is a principle for futures nobody can specify. Post-occupancy evaluation is the habit of going back to see what the future did to a building.
Architects also have the craft most futures teams lack. In speculative design the question of fidelity, how much of a future to build before people can argue with it, is usually the hardest one. Architects answer it every week, moving from sketch to massing model to verified view as the audience changes.
And they already consult. The exhibition, the boards and the Statement of Community Involvement are a ready-made route for putting a future in front of the people who will live with it. None of this needs replacing. It needs pointing at a different date.
The horizon problem: buildings outlive every brief
Most roles that come to speculative design look too near. Architects have the opposite problem. The building outlives every horizon the method recommends, and yet it is briefed, appraised and drawn for the day it opens.
The brief describes the occupants who exist now. The planning policy is the one in force at submission. The render shows today's weather and today's idea of a good afternoon. A scheme handed over in 2030 reaches middle age in the 2050s, probably owned by someone else, used by people who were children at handover, under rules nobody has drafted. A masterplan's phasing plan ends when the last block completes, which is roughly when the place starts to have a life of its own.
The useful move is not to imagine the building's hundredth year. Past twenty-five years or so, people stop being able to picture a future or feel responsible for it. The move is to pick the building's mid-life, twenty years or so after handover, and treat its occupants then as a second client whose brief you have not been given. That keeps the work inside the ten-to-twenty-five-year window where speculation produces decisions.
Where speculative design fits in the RIBA Plan of Work
At the front. The work belongs in Stages 0 to 2 — Strategic Definition, Preparation and Briefing, Concept Design — while the grid, the cores and the aspect are still open. By Stage 4 speculation can only decorate.
| You already use | Its speculative counterpart | What it adds |
|---|---|---|
| The client brief | A second brief, for the occupants twenty years after handover | Tests whether the building is for its users or only its first users |
| Site and context analysis | A signal scan: dated events in climate, work, travel and regulation | Evidence of change local policy has not caught up with |
| Massing and options studies | A future wheel with the building as its subject | Second-order consequences that change the plan, not just the elevation |
| The hero render | The uncomfortable render: the same view on its worst ordinary day, decades out | Shows the building once the brief's assumptions stop holding |
| Consultation boards | Future artefacts: the lobby notice, the tenancy clause, the bus timetable | Lets residents argue with how the place will be run, not how it looks |
| The masterplan phasing plan | A backcast from the place in its third decade | Puts management and institutional change on the construction timeline |
A worked example: the car park that has to become something else
Take a hypothetical practice at Stage 2 on a 600-space multi-storey car park in a Midlands town centre. The council is building it as part of a masterplan that clears surface car parks to free land for housing, and handover is planned for 2029. The hero render shows it at noon, its planted façade full of cars.
The brief. Alongside the client's brief, the team writes a speculative one: shared driverless vehicles (the source of change), the car park and the street it fronts (the domain), 2050 (the horizon), for the council's property team and its highways officers.
Signals. Three dated events, each checkable:
- •On 25 June 2024 Waymo removed the waitlist for its driverless robotaxis in San Francisco, so anyone could hail a car with nobody in the driver's seat.
- •On 3 September 2026 Wayve and Uber launched supervised autonomous rides in London, the first autonomous trips offered to the public in the UK, with a licensed driver on board.
- •In December 2017 Peckham Levels opened in a multi-storey car park on Rye Lane in south London, its storeys converted into studios and workspace let to local businesses.
The third signal matters most to an architect. A car park's second life is already a building type, and whether this one can have it is settled now, in section.
Future wheel. At the centre: most trips into the town centre in 2050 are made in shared driverless vehicles that drop off and leave. The first ring is what transport consultants already say: less parking demand, more kerbside drop-off. The second ring is where the building changes. The fleets still need somewhere to wait, charge and be cleaned overnight, so the lower floors become a depot, busier at three in the morning than at three in the afternoon. The upper decks empty, and the parking income the council had budgeted for falls with them. In the third ring those empty decks are the cheapest floor space in town. If they can be occupied, the town gains a building; if they cannot, it gains a liability the council pays to keep standing.
The concept. The team makes one thing specific enough to argue with:
Upper Deck — "Four floors the cars gave back." The ground and first floors are a charging and cleaning depot for the town's shared fleets, run on a lease from the council. The four decks above have been glazed and let at low rents as workshops, studios and a sixth-form college's engineering annex, reached by a public stair built alongside the old ramp.
The sketch comes in two versions of the same view at eleven on a Tuesday morning in 2050. In the first, Upper Deck exists. In the second, the uncomfortable render, a sign on the stair reads "Levels 3–6 closed": the decks were built as split-level ramps with a car's headroom, so nothing else could use them. Fleet vehicles queue along a pavement that has become a drop-off lane.
Implications. Start with one life: a resident of the flats opposite, woken nightly by the depot the car park became. Then outwards: a high street that loses the footfall parking used to bring; a college sharing a building with a 24-hour vehicle operation. Then the shipwreck: a building rescued for the town by day and turned against its neighbours at night, because the kerb was given to vehicles.
Backcasting. Standing in 2050 and working back, the rungs sort by owner. Some belong to other people: a parking strategy that releases decks, a planning consent for the change of use, a lease to a fleet operator, fire rules for mixed occupation. Those are institutional rungs, held by the people our guide on speculative design for policymakers is written for. The architect's rungs all fall in the next few months, and nearly all are dimensions: flat decks with a separate ramp rather than split levels, floor-to-floor heights with room for a habitable ceiling, floors loaded for people as well as cars, a core sized for pedestrians, a façade that can take glazing later.
What changes this quarter. At Stage 2 the team prices flat decks against split levels, raises the floor-to-floor height, specifies occupancy loading on two decks and moves the core to the street front. It takes both 2050 drawings to the council's property board with the cost difference written underneath. None of those changes is speculative. All of them came from speculation. The car park gets built either way; the question is whether it can ever be anything else.
Where speculative design goes wrong for architects
The ruin render. Asked to show the uncomfortable, architects reach for the flooded atrium and the ivy-covered tower. That is dystopia by default, and it lets everyone off, because nobody designs for the apocalypse. The discomfort that changes a scheme is ordinary: a closed floor on a Tuesday morning, the empty shopfront in year twelve, the bin store after the collection contract changes.
"Future-proof" as sales copy. The work is absorbed into the Design and Access Statement as reassurance rather than as a test. If no drawing changed, it was marketing.
Building the future you imagined. Falling in love with one future and casting it in concrete: the drone pad, the autonomous drop-off bay, the floor flexible for exactly one use. Speculative work on a building should mostly produce loose fit (floor-to-floor heights, grids, flat floors, separate entrances) rather than features. The more specific the future, the less specific the building should be about it.
Speculating about the object instead of the place. The futures that decide whether a building succeeds are rarely about its skin. They are about who holds the key, who pays the service charge and who is allowed in after dark. Our essay on placemaking explains why public space is made by its users as much as its designers. Speculation that stops at the envelope misses most of that.
A first exercise for your next design review
- 1.Take a live project at Stage 1, 2 or 3 and print its hero render.
- 2.Pick a year twenty to twenty-five years after its planned handover and write it on the print.
- 3.Collect three signals relevant to the site, each a dated event with a source. Pin them beside the render.
- 4.On a sheet of trace, draw a future wheel with one signal at the centre and the building as its subject. Go at least two rings out.
- 5.On a layer over the render, redraw the same view on the worst ordinary day of your chosen year. Change the time, the weather, who is there and what is shut.
- 6.Annotate three things the building does badly in that drawing and one it does well. Trace each to a decision being made now.
- 7.Take the cheapest of those decisions, a dimension, an orientation or a door, to the next design team meeting, with both drawings on the table.
Where architects can learn speculative design
Speculative Design Basics is where we teach the whole workflow: 42 self-paced lessons from J. Paul Neeley, assuming no prior experience. For architects the Process chapter matters most, because it covers how the method relates to existing design processes (for you, the Plan of Work) and how to choose a time horizon. The Tools & Methods chapters on signals, future wheels, future prototyping and backcasting map onto the exercise above, and the Case Studies include future mobility. At the scale of the neighbourhood or town, Place Futures applies futures methods to places, with participatory approaches throughout; our essay on future cities sets out the drivers it works with. Neighbouring disciplines are covered in speculative design for every role.
Key takeaways
- •The architectural render is already speculation. It just never shows the building having a bad day.
- •Buildings outlive every horizon the method recommends, so set the work at mid-life, twenty years or so after handover.
- •Render the worst ordinary day, not the ruin: the discomfort that changes a scheme is mundane.
- •Backcasting separates the rungs you own (deck geometry, floor heights, cores) from those held by councils and operators.
- •Work in Stages 0 to 2, and expect the output to be loose fit, not features.
Frequently asked questions
How do architects use speculative design?
Architects use speculative design to test a scheme against the building's middle age rather than its opening day. In practice that means a signal scan for the site, a future wheel with the building as its subject, a render of the same view on its worst ordinary day twenty years after handover, and backcasting to find the decisions, mostly dimensions and orientations, that are still cheap to change at RIBA Stages 0 to 2.
Isn't an architectural render already a kind of speculative design?
A render is already speculation, because it pictures a place that does not exist. But it is uncritical speculation: it shows the one future the scheme is selling, usually a sunny afternoon with mature trees and contented people. Speculative design keeps the render and changes its job. The uncomfortable render shows the same view on an ordinary bad day twenty years on, so the design team can see what the building does when its assumptions stop holding.
How far ahead should an architect look when using speculative design?
Look to the building's mid-life, around twenty to twenty-five years after handover, rather than its end of life. That keeps the work inside the ten-to-twenty-five-year window where speculative design produces decisions, while still reaching well past the brief, the planning policy and the first occupants. Futures set much further out tend to be admired and filed, and a demolition date is too far away for anyone to argue with.
Can speculative design be used in public consultation on a planning application?
Yes, and it often works better than a render. Instead of showing residents how a scheme will look, show them an artefact from how it will be lived in: a notice in the lobby in 2050, a tenancy clause, a drawing of the car park's closed upper floors. People can argue with management, access and use in a way they cannot argue with a façade. Use it while the design can still change.
What is the best way for an architect to learn speculative design?
Start with Speculative Design Basics, the School of Critical Design's self-paced online course of 42 lessons, which assumes no prior experience. Its Process chapter covers how the method relates to existing design processes and how to choose a time horizon, and the Tools & Methods chapters cover signals, future wheels, future prototyping and backcasting. Architects working on neighbourhoods and towns can follow it with Place Futures.