Answer
P50, P75 and P90 label the probability of exceeding an energy estimate under a stated model. P90 is a lower energy threshold with a higher exceedance probability, not 90% of P50 or a production guarantee. In commercial sales, explain the study period, energy boundary and uncertainty assumptions before connecting any estimate to revenue or savings.
A customer comparing commercial solar proposals needs to know what the numbers describe. A P90 figure can sound more certain while representing less energy than P50. Your job is to explain that distinction, retain the study behind it and make clear which assumptions remain unresolved.
This is a sales interpretation and release guide. The P50 versus P90 energy-yield guide covers the technical calculation and uncertainty checks; the P50/P90 definition provides a shorter reference.
Explain the Probability Before the Energy Number
| Label | Exceedance probability under the adopted model | Customer interpretation |
|---|---|---|
| P50 | 50% | The median modeled energy outcome |
| P75 | 75% | A lower energy threshold than P50 for the same distribution |
| P90 | 90% | A lower threshold than P75 for the same distribution |
Higher exceedance probability does not mean higher production. The associated energy thresholds normally decrease as the probability increases. Equal thresholds can occur in some distributions; this table does not prescribe their spacing.
A useful explanation is: “These figures describe different thresholds in the energy study. The P90 threshold has a higher modeled probability of being exceeded than P50. The assessment depends on its weather, equipment, loss and uncertainty assumptions.” Then identify the actual report and period.
PVsyst’s official documentation defines these as exceedance values and describes its normal-distribution method. That is a declared tool assumption, not a universal distribution for every study. Never describe a P-number as an accuracy score or multiply P50 by 0.90 to manufacture P90.
Label the Period and Output Boundary
Before quoting a result, ask the technical reviewer to complete these fields:
| Field | What must be clear before release |
|---|---|
| Energy quantity | kWh or MWh, with the covered period |
| Boundary | DC generation, inverter AC output or delivered energy at the specified meter |
| Study horizon | Single-year energy, multiyear average or another explicitly defined statistic |
| Project state | Equipment/layout revision and model/run identifier |
| Weather basis | Dataset/file, covered years or typical-year construction |
| Uncertainty scope | Included components, excluded effects, dependencies and method |
| Reviewer | Person responsible for interpreting and approving the study |
A single-year P90 and a P90 of a multiyear average address different events. Do not compare them as though their only difference were the software vendor. Averaging may reduce relevant independent random variation; systematic bias and correlated uncertainty do not automatically disappear. Do not divide every uncertainty term by the square root of the number of years.
A reported annual energy value also does not provide a validated hourly P90 profile. Avoid uniformly scaling every hourly result by an annual P90/P50 ratio to create one. The vendor documentation warns against that shortcut. For a tariff or demand-charge assessment, request a suitable time-series method and state its limitations.
A Small Example for the Sales Conversation
Assume an illustrative normal model with a median/mean annual energy of 1,000 MWh and standard deviation of 50 MWh. These are invented teaching inputs, not an observed project, regional benchmark or SurgePV output.
| Scenario | Calculation using rounded normal quantiles | Illustrative energy |
|---|---|---|
| P50 | 1,000 | 1,000 MWh |
| P75 | 1,000 − 0.6745 × 50 | About 966 MWh |
| P90 | 1,000 − 1.2816 × 50 | About 936 MWh |
The P90 is about 93.6% of P50 in this example. It was not created by choosing a standard 10–15% haircut. A different uncertainty model would change the gap. Nor should the salesperson linearly interpolate P75 between two supplied values: obtain the result from the adopted method.
Keep this example separate from project documents. For an actual proposal, quote the approved assessment’s values and rounding, not a number calculated from an assumed “typical” uncertainty. The laboratory report on uncertainty in PV energy estimates discusses resource variability, performance-model uncertainty and reliability considerations. The scope of the project’s study should be explicit, especially where material effects were not modeled.
Connect Energy to Finance Without Turning It into a Guarantee
An energy scenario is an input to a financial model. It does not, by itself, establish bill savings, revenue, debt-service coverage or lender approval. The financial assessment also needs the appropriate prices, load/export treatment, operating assumptions and finance terms.
A total annual energy number alone cannot establish self-consumption or peak-demand savings. Those require compatible load and generation profiles, tariff rules and the relevant meter boundary. Likewise, a contractual performance guarantee needs its own agreed measurement, exclusions, remedies and responsibilities; a P90 label does not create that contract.
Use the scenario requested by the actual customer, lender or investment reviewer. Preserve its name and study period in the generation and financial modeling evaluation rather than silently replacing a base case with a “conservative” figure. Confirm the available reporting method and released product scope with the vendor; this article does not claim that SurgePV automatically calculates or certifies P-values.
Resolve These Questions Before Sending the Proposal
| Customer question | Practical response or action |
|---|---|
| “Will we definitely receive the P90 output?” | Explain model dependence and separate contractual guarantees. Escalate requests for commitments. |
| “Why is another supplier’s P90 higher?” | Align layout, equipment, weather, losses, boundary, period and uncertainty scope before comparing. |
| “Does a smaller P50–P90 gap prove a better model?” | Check whether the study uses better evidence or simply omits uncertainty. The gap alone is insufficient. |
| “Can you give us monthly P90 savings?” | Request a qualified method linking time-series energy, load and tariff. Do not scale a yearly probability number into monthly guarantees. |
| “What changed since the previous proposal?” | Identify the revised input, successor run and affected energy/financial outputs. Retain the previous release. |
Keep the Evidence with the Released Version
Retain the approved report, inputs and assumptions, open items, run identifier and reviewer with the customer-facing proposal. Use the production-estimate assumption register for input ownership and revision history. A pending weather or loss input should remain visible rather than becoming a silent sales assumption.
Check that layout, equipment and reported energy refer to the same project revision. Then check that the financial scenario and proposal quote that released energy result. Assign an owner to unresolved material questions and escalate technical interpretation to the qualified reviewer. If an input changes, preserve the old record, rerun the affected assessment and identify which customer artifacts need correction.
When evaluating software, book a SurgePV demonstration to inspect the actual model inputs, outputs and proposal workflow. Bring the report fields your team needs; confirm product scope rather than presuming that a polished proposal includes an independently reviewed uncertainty assessment.
Frequently Asked Questions
Is P90 equal to 90% of P50?
No. P90 describes exceedance probability, not a fixed energy discount. Its relationship to P50 depends on the model, uncertainty assumptions and study period.
Does P90 guarantee production in nine of the next ten years?
No. A probability statement under a model does not promise a particular sequence of future results or create a contractual production guarantee.
Can a sales team calculate P75 from P50 and P90?
Not without knowing the adopted distribution and method. Ask the technical reviewer for a consistent P75 result rather than interpolating between report values.
What should accompany a P-value in a proposal?
Include energy units and boundary, study period, model/run identity, uncertainty scope, source report and reviewer. Keep financial assumptions and any contractual commitments separate.
Sources
Primary research and reference material used for this desk-research article.
Where this fits
This article is part of SurgePV's Solar Business & Operations hub, which works through the topic from first principles to the decisions a project team actually has to make.
