Executive summary
Sunseam maintains a ranked dataset of 137,849 commercial rooftops across the ten regions of England and Wales, re-scored every week from official public data. This paper sets out five structural findings from that dataset. Each one changes how an installer, a building owner or a landlord should think about commercial rooftop solar.
The market is still almost untouched. Nine in ten roofs in our dataset have no recorded solar installation, and industry estimates put rooftop coverage below 5% nationally.
Ownership, not demand, is the practical bottleneck. About a third of the roofs we can trace sit on a registered title shared with other units, which means a landlord, not the occupier, controls the decision.
Shared ownership is leverage, not friction. The average shared title in our data covers roughly six units, and the largest covers 80. One conversation can open several roofs.
The energy rating clock is running. Nearly two thirds of the stock sits below band B, the level government has targeted for large commercial buildings by around 2031.
The industry undersizes its own opportunity. Measuring actual roof geometry raised installable capacity estimates by about a fifth against the standard floor area rule of thumb.
Aggregated across the dataset, we estimate about 40 million kWp of installable capacity carrying roughly £6 billion a year in combined bill savings and export income. All figures are estimates, and the methodology and its limits are set out at the end of this paper.
Finding 1: Nine in ten roofs, nothing on them
Of the 137,849 roofs we rank, about 90% have no solar installation recorded against the building, and a further 10% carry a partial system with meaningful roof area still available. Buildings already fitted to capacity are excluded from the dataset entirely, so the figure describes genuine remaining opportunity, not a market that has quietly filled up.
The economics are not the obstacle. Commercial electricity prices around 28p per kWh against typical payback of four to nine years make the case for daytime-heavy users comfortably. What the coverage number tells us is that the constraint sits earlier in the funnel: finding the right buildings, and reaching the person who can say yes. The rest of this paper is about those two constraints.
Finding 2: The ownership bottleneck, quantified
The feedback we hear most from installers is some version of the same observation: most commercial occupiers lease their premises, so the tenant who wants solar cannot authorise it alone. That is true, and until now nobody could say how much of the market it actually affects.
We traced each roof in our dataset to the registered title it sits on. Around four in five roofs can be traced. Of those, about a third sit on a title shared with at least one other unit, which is the signature of a multi-let estate: one owner, several occupiers, one decision maker for all of them.
Share of roofs on shared titles by region, August 2026. Figures are approximate and move as the dataset refreshes.
| Region | Roofs ranked | Traced to a title | Shared title (of traced) |
|---|---|---|---|
| North East | 6,452 | 82% | 43% |
| South East | 20,274 | 81% | 41% |
| London | 19,569 | 76% | 40% |
| North West | 18,171 | 79% | 40% |
| West Midlands | 15,236 | 79% | 38% |
| East of England | 13,673 | 81% | 37% |
| Wales | 6,510 | 83% | 36% |
| Yorkshire and the Humber | 13,882 | 81% | 35% |
| East Midlands | 11,810 | 80% | 33% |
| South West | 12,272 | 81% | 32% |
Two things stand out. First, the pattern is national. No region falls below 32% and none rises above 43%, so this is a structural feature of commercial property, not a quirk of one market. Second, the shared titles are not small. The average shared title in our data covers 5.9 units, and the largest single title we have found covers 80.
The industry reads leasehold as friction. The data says it is leverage. If one landlord controls six roofs, that is not six dead ends. It is one conversation worth six installations, and the portfolio owner has a stronger commercial case than any single tenant: one procurement, one contractor relationship, and an energy rating uplift across the whole estate at once.
One honest limitation. A building on its own title can still be leased, because occupational leases are not visible in the data we use. What we can state with confidence is where a landlord is definitely in the picture, and that alone covers a third of the traced market.
Finding 3: The energy rating cliff
The minimum energy rating to let commercial property in England and Wales is band E today. Government has targeted band B for large non-domestic buildings, those over 1,000 square metres, by around 2031. The earlier 2027 band C milestone was dropped in 2026, which has led some owners to relax. The data suggests they should not.
Taking the East Midlands as a representative region: of 11,810 ranked roofs, only 37% sit at band B or above. Nearly two thirds of the stock has work to do within roughly five years, and around a third sits at band D or below, where the gap to B is a serious capital project rather than an operational tweak.
Rooftop solar is one of the few interventions that improves a building's assessed rating while directly cutting the occupier's costs. For the multi-let estates identified in Finding 2, the two problems compound in the landlord's favour: a single solar programme across a shared-title estate addresses the rating exposure of every unit at once.
Finding 4: Grid capacity is a siting problem, not a background condition
Network capacity is usually discussed as a national abstraction. At building level it is anything but. In the East Midlands, of the roofs where we hold a local grid signal, roughly 30% sit in areas with headroom, 30% in moderate areas, and 40% in areas we classify as constrained.
The practical consequence: two identical warehouses a few miles apart can face materially different connection journeys, and the difference is knowable before anyone drives to either. For installers this belongs in qualification, not discovery. A large roof in a constrained area may still be a fine project, but it is a different sales conversation and a different timeline, and it should be priced as such from the first call.
Finding 5: The industry undersizes its own opportunity
The standard shortcut for estimating system size is to work from a building's floor area. It is fast, and it is systematically wrong in a conservative direction: floor area misses multi-storey geometry, and it misses the difference between a building's footprint and its usable roof.
We re-sized our dataset using measured roof geometry wherever a reliable footprint could be matched, currently around two thirds of roofs. Among large buildings with a clearly separable roof outline, measuring rather than assuming raised the installable capacity estimate by about a fifth. The floor area rule of thumb is a reasonable approximation, but it is consistently low, and it is lowest on the buildings most worth quoting first. In the East Midlands the median estimated system moved from 136 kWp to 177 kWp.
The implication runs beyond our dataset. Any pipeline, portfolio valuation or national potential study built on floor area rules of thumb is likely understating the commercial rooftop resource by roughly a quarter.
What this means in practice
For installers
Qualification order matters more than list size. The highest-value approach in this data is a large-roof, daytime-heavy building on a shared title in a headroom area: one conversation, several roofs, a motivated owner and a clean connection. Working ranked data beats canvassing not because canvassing cannot work, but because a third of the market cannot be won at the door you are knocking on.
For building owners and occupiers
If you lease, the path to solar runs through your landlord, and the strongest argument you can hand them is the estate-wide one: rating uplift and energy income across every unit they hold, not just yours. If you own, nearly two thirds of your peers have rating work to do before roughly 2031, and contractors' calendars will not stay empty.
For landlords and asset managers
Shared-title estates are where this market's leverage concentrates, and the data says that is a third of everything traceable. A portfolio solar programme is one of the few capital projects that simultaneously addresses regulatory exposure, tenant retention and income. The owners who move before the 2031 horizon set terms; the ones who move after it accept them.
Methodology, limits and attribution
The dataset covers 137,849 commercial rooftops across the ten regions of England and Wales, assembled from official public data on buildings, energy performance, land registration and network capacity, and re-scored weekly. Scotland and Northern Ireland operate separate registers and are outside the dataset. Capacity and savings figures are modelled estimates built on standard industry assumptions, stated as estimates throughout, and capped so that modelled bill savings never exceed a building's own consumption. Roughly four in five roofs are traced to a registered title; untraced roofs are counted in neither the shared nor the sole category. Title extents record freehold boundaries and do not contain occupational leases, so no claim is made about the tenure of any individual occupier. Ownership figures describe title structure only and identify no proprietor.
Method note, August 2026. Adjoining commercial units frequently share a single mapped roof outline. Where that happens, capacity is now attributed so that a given roof contributes once to regional and national totals. This reduces our published aggregate and does not affect roof counts, energy ratings or ownership figures.
Figures are Sunseam estimates, not guarantees. They move as the dataset refreshes, and this paper states them as at August 2026.
About Sunseam
Sunseam is location intelligence for commercial rooftop solar: every qualifying commercial roof in a region, ranked by a proprietary opportunity score, with the estimated system size and annual benefit for each, and a built-in pipeline to work them. It is free to start at sunseam.co.uk.