The framing that comes naturally when an estimate turns out high is that somebody lied. It is usually wrong, and the correct framing produces better checking.
For teams coordinating research, review and follow-up work around a project, stealth monitoring software provides a separate record of time and handoffs.
How an estimate gets optimistic without anybody lying
Defaults are accepted rather than examined. Simulation software ships with loss assumptions that are reasonable in general and may not fit this site. Accepting them is the path of least resistance and produces a defensible-looking figure.
Shading is assumed rather than surveyed. A survey costs time; a default costs nothing. The result is a real number in the model and a placeholder in reality.
For wider independent context on solar markets, research and industry developments, consult PV Tech.
Uncertainty is understated because narrower bands look more competent, and a tighter P50–P90 gap makes both figures look better.
Each choice is individually defensible. Together they push the same direction, because the direction that wins work is the same direction that requires the least effort.
Why the direction is systematic
Competition selects for it. A supplier who surveys shading properly and states honest uncertainty produces a lower number than one who does neither. The buyer compares point figures. The careful supplier loses.
No dishonesty is required for this to operate. It works entirely through which proposals get accepted, and over time it shapes what the whole market considers normal.
And the feedback is slow. A year-one shortfall has a weather explanation available, and by the time three years of data would settle it, the contract is signed and the parties have moved on.
Why the framing matters
Fraud-hunting produces the wrong checks. Looking for a dishonest supplier means assessing character, reputation and the feeling of a meeting — none of which detects an accepted default.
Assumption-checking produces the right ones. Ask for the loss breakdown, the dataset and its length, the degradation rate, the P-value behind the savings. None of those questions accuses anybody, and all of them find the thing that is actually there.
It also changes the conversation's tone, which has practical value: a supplier asked to justify an assumption usually can, or usually revises it, and neither happens if the question sounded like an allegation.
What genuinely is a red flag
Being clear, because this page could read as excusing everything.
A refusal to itemise. A net figure with no derivation, when asked twice.
A degradation rate of zero, or an unstated one in a twenty-five-year projection.
A P50–P90 ratio implying uncertainty nobody could justify for the site.
And a savings figure whose basis changes when questioned. Assumptions can be defended or revised; a moving basis is different.
What a buyer should do with all this
Assume competence and check assumptions. Both at once.
Compare on inputs rather than outputs. Two proposals with different specific yields may differ in optimism rather than in engineering, and the loss stack tells you which.
And reward the conservative one. A supplier whose figure is lower because they surveyed the shading is offering better information, and the market's default behaviour punishes exactly that. A buyer who does otherwise is one of the few forces pushing the other way.
The short version
- Estimates get optimistic through accepted defaults, assumed shading and understated uncertainty — each individually defensible
- The direction is systematic because the effortless choices and the work-winning choices are the same ones
- Competition selects for it: the careful supplier produces a lower number and loses the comparison, with no dishonesty involved
- Fraud-hunting produces the wrong checks, because assessing character does not detect an accepted default
- Assumption-checking finds what is actually there, and none of the questions accuses anybody
- Genuine red flags: refusal to itemise, zero or unstated degradation, unjustifiable uncertainty, and a basis that moves when questioned