After-tax IRR modelling: what installer brochures get wrong

Commercial solar installer marketing routinely quotes 25-year IRRs in the 25–35% range. Our modelling of the same projects, with the same capex and demand assumptions, produces 14–20%. That's a meaningful gap that confuses customers — and the gap reflects four specific methodological errors that recur across most installer financial models.

Error 1: Tariff inflation assumption

Most installer models assume 5–7% annual electricity tariff inflation over 25 years. That assumption was tenable in 2018–2022 when wholesale prices rose sharply. It's not tenable now.

Forward wholesale prices into 2030 trade at £75–£90/MWh real (Q1 2026). The forward curve is broadly flat in real terms. Combined with declining policy-cost components (peak likely passed), the net real-terms tariff trajectory is approximately flat — call it 0.5–2% real over 25 years rather than 5–7%.

The compounding effect over 25 years is enormous. At 6% inflation, year-25 saving is 4.3× year-1. At 1.5% inflation, year-25 saving is 1.45× year-1. The lifetime IRR difference is 5–8 percentage points just from this single assumption.

Error 2: Inverter replacement cost

Inverters are the single largest mid-life cost on commercial solar. They typically need replacement at year 10–13 — sometimes pulled forward by warranty failure, sometimes deferred slightly by careful maintenance. The replacement cost is £80–£120 per kWp installed.

Many installer models simply omit this cost. On a 250 kWp system, that's £25,000 of present-value cost (discounted from year 12 at typical rates) excluded from the lifetime model. Including it cuts lifetime IRR by 1.5–2.5 percentage points.

Honest models build a year-12 inverter replacement reserve into the project financial assumptions and depreciate it forward through the 25-year window.

Error 3: Self-consumption assumption

Many installer models assume 90%+ self-consumption regardless of demand profile. Real numbers vary 60–90% depending on site type, time-of-use pattern, and seasonality. A 10-percentage-point shift in self-consumption (from 90% to 80%) drops year-1 saving by approximately 6% (because the shifted kWh moves from ~17p avoided cost to ~7p export tariff — a 10p value differential).

Honest models pull half-hourly demand data and compute actual self-consumption against that data, not against an assumed percentage. Where half-hourly data isn't available, conservative assumptions (75% self-consumption) are appropriate rather than optimistic ones (90%).

Error 4: O&M costs

Operating and maintenance costs run £8–£12/kWp/year for typical commercial solar — monitoring, periodic cleaning, minor repairs, insurance. Many installer models either omit O&M entirely or assume a token £2–£3/kWp/year.

On a 500 kWp system, the difference between £2/kWp and £10/kWp annual O&M is £4,000/year. Over 25 years that's £100,000 of cumulative cost — roughly 25% of capex on a £400k system. Excluding it materially overstates lifetime returns.

Putting the errors together: a worked re-modelling

Take a 250 kWp system at £200k capex. Installer model: 90% self-consumption × 22p × 6% tariff inflation × no inverter replacement × £2/kWp O&M = lifetime IRR of approximately 28%.

Honest model: 78% self-consumption × 22p × 1.5% real tariff inflation × £25k inverter replacement at year 12 × £10/kWp O&M = lifetime IRR of approximately 16%.

Same project, 12 percentage points of lifetime IRR difference. The honest 16% is still a strong return — but it's a different conversation than the headline 28%.

How to read installer financial models

Practical questions to ask any installer financial proposal:

• Tariff inflation assumption? Should be 1–2% real, not 5–7% nominal. Anything above 3% real warrants challenge.
• Self-consumption percentage? Should be derived from half-hourly data if available, conservatively assumed (75%) if not. 90%+ assumptions warrant challenge.
• Inverter replacement? Should explicitly model year-10 to year-13 replacement at £80–£120/kWp. Models without inverter replacement cost are systematically optimistic.
• O&M cost? Should be £8–£12/kWp/year. Anything below £5/kWp/year warrants challenge.
• Tax allowance treatment? Should explicitly model FYA timing against your accounting year-end and AIA headroom against your wider capex position.

A model that fails on three or more of these questions is unlikely to be giving you the actual project economics. A model that handles all five carefully — even if it gives a lower headline IRR — is the one to trust on commitment decisions.