A rooftop photovoltaic proposal arrives with a payback figure attached. Seven years, say. The figure is almost never dishonest. It is the output of a calculation with perhaps a dozen inputs, each chosen reasonably, and each chosen at the favourable end of a reasonable range.
Move seven of those inputs to the other end of their reasonable range and the same installation pays back in eleven years. Nothing was falsified. The reader simply never saw where the dials were set.
This explains the seven inputs that matter most, roughly how far each one can move the answer, and what to ask so that a proposal becomes checkable.
The problem with one number
Payback is a ratio of a cost you know to a benefit you are forecasting over twenty years. The cost is firm; almost everything else is an assumption. Presenting the result as a single number implies a precision the method cannot support.
We publish ranges instead, with the inputs that drive the spread named. A range of 7.4 to 11.2 years is a more honest and more useful answer than 8.3 — particularly if your decision threshold sits inside the range, which is exactly when you need to know.
Assumptions behind this explainer
- A commercial or industrial rooftop installation in Bulgaria, self-consumption led, without subsidy or grant support.
- Crystalline silicon modules with a string or central inverter; no storage. Storage changes the analysis materially and is out of scope.
- All figures are illustrative, chosen to show the arithmetic and the direction of sensitivity. They are not a quotation, a benchmark, or a claim about current market prices.
- Simple payback on undiscounted cash flow unless stated. Discounted measures are discussed under assumption 6.
- Tariff and regulatory treatment change; verify current arrangements before relying on any of this.
The seven assumptions
1. Specific yield
Expressed as kWh per kWp per year. In Bulgaria a fixed, roof-mounted, unshaded array typically falls somewhere around 1,100–1,350 kWh/kWp/year, varying with latitude, orientation, tilt and local climate. Southern sites outperform Sofia; a flat-roof east–west layout gives up yield relative to south-facing at optimal tilt, in exchange for a flatter generation profile that often self-consumes better.
Ask which figure was used and where it came from. The European Commission's PVGIS tool gives a free, independent, location-specific estimate, and takes minutes. If a proposal's yield sits above PVGIS for the same location and geometry, that difference is the first thing to query.
2. Self-consumption rate
The share of generation you use on site rather than exporting. This matters more than anything else in the calculation, because self-consumed energy is worth what you would have paid to buy it, while exported energy is worth whatever the export arrangement pays — usually much less.
A site running weekday daytime shifts might self-consume 70–90%. A site with weekend shutdowns and a summer holiday may be at 40–55%. That difference alone can move payback by several years on an otherwise identical installation.
It is also the assumption most often estimated rather than calculated. Doing it properly needs your interval consumption data overlaid on a modelled generation profile — which is why metering data quality determines whether this number is knowable at all.
3. Degradation
Modules lose output over time: commonly an initial drop in the first year, then roughly 0.4–0.6% per year for modern crystalline silicon. Over twenty years the difference between assuming 0.3% and 0.7% compounds into several percent of lifetime generation.
Small relative to self-consumption, but it acts on every year of the model, and it is sometimes omitted entirely — which quietly overstates the later years.
4. Tariff trajectory
The benefit of self-consumed generation equals the price you avoid paying. So the payback depends on a twenty-year electricity price forecast, whether or not the proposal admits to containing one.
Assume 5% annual escalation and the case looks strong. Assume flat prices and it weakens considerably. Neither is knowable. What matters is that the assumption is stated, and that you see the result under a flat-price case as well — that is the version that tells you whether the investment stands on its own.
The structure matters as much as the level. If a significant part of your bill is a capacity or demand charge rather than an energy charge, solar reduces the energy component but may barely touch the rest.
5. Operating costs and inverter replacement
Ongoing costs are modest but not zero: cleaning, inspection, monitoring, insurance. The item most often missing is inverter replacement. Inverters typically do not last the life of the modules, so a twenty-year model should normally contain at least one replacement. A model that shows no capital expenditure after year zero is describing a system that will not be running in year twenty.
6. Discount rate
Simple payback treats a euro saved in year fifteen as equal to a euro saved today. For a twenty-year asset that is a significant distortion.
A discounted analysis at your actual cost of capital lengthens payback and is the honest basis for comparing this project against other uses of the same money. Ask for net present value and internal rate of return alongside simple payback, and ask what discount rate was used. If the answer is that no discounting was applied, you know how to read the headline figure.
7. Grid connection and regulatory treatment
Connection terms, any export limitation, metering arrangements and the treatment of exported energy all sit upstream of the entire calculation. An export limit that curtails generation at exactly the times you produce most will change the yield in assumption 1 and the self-consumption in assumption 2 simultaneously.
This is also the assumption with the shortest shelf life. Regulatory arrangements are revised; a model built on last year's terms may not describe this year's.
How much each one moves it
Indicative only, and illustrative — the point is the ranking and the direction, not the magnitudes, which depend on your own base case.
| Assumption | Plausible range | Effect on payback |
|---|---|---|
| Self-consumption rate | 45% → 85% | Very large |
| Tariff escalation | 0% → 5% per year | Large |
| Specific yield | 1,100 → 1,350 kWh/kWp | Moderate |
| Discount rate | 0% → 8% | Moderate |
| Inverter replacement | Omitted → included once | Moderate |
| Degradation | 0.3% → 0.7% per year | Small |
| Operating costs | Low → high | Small |
The ordering carries the practical message. Effort spent establishing your real self-consumption rate from metering data is worth more than precision on any other input, and it is the one input you can actually measure rather than assume.
A worked range
Cumulative cash position for one illustrative installation, under a favourable and an unfavourable set of assumptions. Same array, same cost, same roof.
Both lines are defensible. Both would appear in a competent proposal. Only one usually does.
Reading a proposal
Five questions, all reasonable to ask, none adversarial:
- What specific yield did you assume, and how does it compare with PVGIS for this roof?
- What self-consumption rate, and was it calculated from our interval data or estimated?
- What tariff escalation — and what does the payback become at 0%?
- Does the model include inverter replacement, and in which year?
- Is this simple or discounted payback, and at what rate?
A supplier who answers these readily is one worth dealing with; the questions are ordinary engineering hygiene. Reluctance to answer question 2 or 3 is the most informative outcome you can get, and it costs nothing to ask.
Limits of this explainer
Use this to interrogate a proposal, not to replace one. Nothing here substitutes for a site assessment by someone who can inspect the roof, the electrical infrastructure and the connection terms, and it is not a design method.
Storage, grants and subsidy schemes, tax treatment and carbon reporting all sit outside its scope. Any one of them can change the decision on its own, whatever the payback figure says.
Found an error? Tell us and show your working — we correct in place and record what changed. Contact us.