Commercial rooftop solar economics in the UK are, in large part, a load-matching problem. The array produces most of its output during daylight hours. If the building uses that power on site, you avoid buying it from the grid at your import tariff. If you do not, it exports — and export is almost always worth less per kWh.
Self-consumption and export are the two pipes every generated kWh flows through. Feasibility quality often comes down to how accurately those pipes are sized for your building — not how aggressively the sales deck rounds the percentage.
Defining self-consumption and export
Self-consumption (sometimes called self-use or on-site utilisation) is generation consumed behind the meter in real time. It reduces import on the same half-hour settlement period, subject to metering arrangements.
Export is generation not used on site and sent to the grid. Revenue depends on a Smart Export Guarantee tariff, a corporate PPA, or market arrangements — rarely matching the full import unit rate.
Self-consumption ratio is self-consumed kWh divided by total generation, usually expressed annually. It is the single most influential assumption in many commercial solar payback calculations.
Behind-the-meter solar on commercial sites is usually metered on a half-hourly basis under P272-style arrangements for larger supplies. Feasibility does not replace metering design, but the financial model should be consistent with how import and export will be settled once live.
The commercial solar financials hub places load match in context alongside capex and generation.
Why load profile matters more than sector labels
“Warehouse” or “office” tells you little about self-consumption. What matters is when the building draws power relative to when the roof produces.
A distribution centre running conveyors Monday to Saturday has a different profile from a spec office occupied weekdays only. A cold store with continuous refrigeration may consume through the night — but solar still only generates by day, so overlap depends on daytime process load.
Half-hourly HH electricity data is the gold standard for commercial feasibility. Where it is unavailable, feasibility should use a declared proxy — for example, estimated annual kWh with a stated occupancy pattern — and flag uncertainty in sensitivity.
Stage1Energy’s methodology explains how consumption data feeds the financial model in the site assessment dossier.
Valuing self-consumed energy
Avoided import is valued at the marginal unit rate you would otherwise pay — including time-of-use bands if your tariff has them. Peak-heavy load with peak-overlapping solar is more valuable than flat load with the same annual kWh.
Some corporates also attribute non-energy benefits: reduced exposure to volatile wholesale prices, REGO claims, or carbon reporting improvements. Those may appear in the business case narrative but should be separated from core cashflow unless quantified conservatively.
Landlord–tenant arrangements complicate who captures self-consumption value. If the tenant pays the bill, the landlord needs a lease mechanism to share benefit. That structure belongs in building a commercial solar business case.
Valuing export
Export revenue in feasibility models typically uses:
- A named SEG tariff from a licensed supplier
- A stated p/kWh PPA offer
- A conservative wholesale-linked proxy
Overstating export is one of the most common ways to shorten apparent payback. Over 25 years, it also inflates NPV — see commercial solar NPV explained.
Export limits imposed by the DNO can cap how much flows to the grid even when on-site demand is low. Grid screening at commercial solar feasibility stage prevents export-heavy business cases from ignoring connection reality.
Improving self-consumption without guessing
Options to increase on-site use include:
- Load shifting — moving discretionary demand into solar hours where operationally feasible
- Electrification — new daytime loads such as EV charging depots aligned with array output
- Battery storage — storing midday export for evening use, at additional capex and round-trip losses
- Tenant engagement — aligning occupancy or process schedules with generation on multi-let assets
Feasibility can scenario these; the base case should reflect current or committed reality, not best-case behaviour never agreed with operations.
Modelling the split in a 25-year cashflow
Annual cashflow applies a self-consumption ratio to degraded year-n generation, values self-consumed kWh at import rates and exported kWh at export rates, and nets operating costs. The ratio may be held constant or varied by scenario.
The mechanics sit in how a 25-year solar cashflow model works. Small changes in ratio move payback and IRR materially — worth a sensitivity table in every board paper.
Red flags in self-consumption assumptions
Challenge the model if you see:
- 90%+ self-consumption on a lightly occupied building with no data
- Export valued at the same rate as import
- No mention of metering or DNO export limits
- Weekend-zero load assumed on a site that runs six or seven days
Independent feasibility should state data sources. The example report shows how consumption and generation are reconciled in a worked UK dossier.
Screening vs full assessment
Free screening in three working days gives a verdict on whether load match and roof resource make deeper modelling worthwhile — without committing to the full financial workbook.
The site assessment — 29 pages, £1,250 per site, five working days — models self-consumption and export with transparent assumptions inside the full cashflow, NPV, and IRR output.
Self-consumption is not a detail in the footnotes. On UK commercial rooftops, it is usually the story the financials are telling.
If you only have annual kWh today, order half-hourly data from your supplier before the board meeting. The cost is modest compared with structural survey spend — and it routinely shifts the case by a full percentage point of self-consumption or more.
To model these metrics for a named UK commercial roof, see solar feasibility study cost in the UK or start with free screening.