A yield estimate receives intense attention for a few weeks and none afterwards. Nobody goes back in year five and asks whether it was right.
For teams coordinating research, review and follow-up work around a project, workforce optimization software provides a separate record of time and handoffs.
That missing loop is why optimistic assumptions persist, and it is the cheapest thing in this whole subject to fix.
Why nobody does it
The people who would benefit have moved on. The buyer made the decision; the supplier delivered; the plant is somebody else's operational responsibility.
For wider independent context on solar markets, research and industry developments, consult IEA PVPS.
It requires data nobody kept. Actual irradiance for each year, against actual production, against the original model — and the model itself is frequently in an email nobody can find.
A single year proves nothing and everybody knows it, so the check is deferred to a point at which it would be meaningful and never scheduled.
And the answer is awkward. If the estimate was optimistic, remedies depend on contract terms that usually will not help, and the finding produces discomfort rather than money.
What the missing loop costs
Suppliers get no correction. An assessor whose estimates run high finds out through nothing — no client returns with three years of data — so the assumptions that produced the high figure survive into the next proposal.
Buyers repeat the mistake. The second installation is evaluated with the same inability to distinguish a careful estimate from a generous one.
And the market's benchmarks drift. What counts as a normal specific yield is set partly by what proposals claim, and if claims are never tested, the claimed range and the achieved range separate quietly.
The check, and what it costs
Three years of four numbers. Production, irradiance, performance ratio, availability — which you should be recording anyway.
Compare cumulative actual production against cumulative P50 over the same period.
Roughly at or above P50 across three years: the estimate was sound.
Consistently below P50 with design-level performance ratio: the model was optimistic and the plant is fine.
Consistently below with low PR: the plant, not the model.
Total effort: an hour, once, in year three. It is the highest-value hour available to anybody operating a solar plant, and almost nobody spends it.
What to do with the answer
Tell the supplier either way.
If the estimate held, that is worth knowing to them and worth saying — and it makes you a reference with something specific behind it.
If it did not, the information is more valuable than any remedy you are likely to obtain. A supplier told that their model ran eight per cent high across three years, with the irradiance data to support it, has been given something they cannot get any other way.
Some will act on it. That is the loop closing, and it requires one party to bother.
Why this site cares
Because the publisher is a supplier, and a supplier with no feedback is a supplier drifting.
The industry's estimates would improve if buyers routinely fed back three-year results, and no regulation or standard is going to produce that. It requires owners to keep four numbers and spend an hour.
That is an unglamorous ask and it is the one intervention available to the party who actually paid.
The short version
- A yield estimate is scrutinised once before purchase and never tested against the years that follow
- Nobody does it because the interested parties have moved on, the data was not kept, one year proves nothing, and the answer is awkward
- The cost: suppliers get no correction, buyers repeat the mistake, and the market's benchmarks drift away from achieved performance
- The check is three years of four numbers against cumulative P50, and it takes an hour in year three
- Below P50 with design-level performance ratio means the model; below with low PR means the plant
- Tell the supplier either way — a three-year result with irradiance data is information they cannot obtain any other way