A kilowatt-hour you use on site displaces one you would have bought at retail. A kilowatt-hour you export earns whatever the export arrangement pays.
Commercial review also depends on how people organise decisions and handoffs; for a separate workplace-management perspective, see this resource.
Those are frequently different by a factor of three or four, which makes the split between them more consequential than the yield figure itself.
Reviewed August 9, 2026. Tariff structures and export arrangements vary by country and change; verify current terms locally.
For broader context on solar procurement, equipment markets and current industry practice, consult ScienceDirect.
Why the values differ
A retail tariff includes more than energy. Network charges, levies and taxes that scale with consumption are avoided along with the energy itself, and that is why self-consumed output is worth the full displaced rate.
An export payment is usually energy only, at a wholesale-linked or administratively set rate.
And some parts of a bill do not scale at all — fixed connection charges continue regardless, so the avoided amount is less than the total bill divided by kilowatt-hours.
Work out your own displaced rate from an actual bill, separating what scales from what does not. It is the input the whole calculation rests on and it is specific to you.
What determines the split
The match between your load and the generation curve. Solar peaks at midday; a business consuming at midday captures most of it and a residential pattern with evening load does not.
System size relative to load. Oversizing raises total production and lowers the self-consumed share, which is why payback favours smaller systems.
Seasonality. A summer peak in generation against a winter peak in consumption exports in one season and imports in the other.
And storage, which shifts generation into later hours at a capital cost and its own round-trip loss.
The number to ask for
"What self-consumption share does this proposal assume, and how was it derived?"
From interval data — your actual half-hourly or hourly consumption for a year, matched against modelled generation. This is the good answer and it requires data you can usually obtain from your supplier.
From a monthly bill total — a much weaker derivation, because monthly totals cannot show whether consumption coincided with generation.
Or from a typical profile for your sector — a default, and a default is a placeholder.
The difference between these three methods can move the answer by twenty points or more, and twenty points moves the investment case substantially.
Sizing against the split
The largest system that fits is rarely the best investment, and neither is the smallest.
Where export is worth little, sizing close to your daytime base load maximises value per unit invested.
Where export is well paid, a larger system can be better despite the lower self-consumed share.
And where export is capped or curtailed, that constraint belongs in the model rather than in a footnote.
Ask for the calculation at two or three sizes. The supplier's model already supports it, and the shape of the result tells you where the sensible point is for your situation rather than for a generic one.
What can change afterwards
Worth planning for, because the asset outlasts the arrangement.
Export rates change. An arrangement set today may not hold for twenty-five years, and a case that depends heavily on export income is exposed to that.
Your consumption changes. A process added or removed alters the split materially.
And tariff structures change — a shift in how network charges are levied can alter the value of self-consumption without the energy price moving at all.
A case that works only under current terms is worth identifying as such before it is signed.
The short version
- A self-consumed kilowatt-hour displaces a full retail rate; an exported one earns an energy-only rate, frequently three or four times less
- Work out your own displaced rate from an actual bill, separating charges that scale from those that do not
- The split depends on load match, system size relative to load, seasonality and storage
- Ask how the assumed share was derived: interval data is good, monthly totals are weak, a sector default is a placeholder
- The derivation method can move the answer by twenty points or more, which moves the investment case substantially
- Ask for the calculation at two or three sizes, and identify whether the case depends on export terms that may not last