Answer first: obtain the run and its inputs, not just the annual headline
Before treating a PVWatts estimate as a home-specific production forecast, ask the provider for the dated model run, version, actual inputs, weather-data provenance and limitations. A forecast can be useful without being a performance promise, an equipment diagnosis or a savings calculation. A panel count and an annual energy total alone do not show whether the model represents your roof and the system being proposed or already installed.
The official current PVWatts V8 documentation identifies separate model inputs and response fields. Its loss help explains that some effects are modelled separately from the general loss allowance. An older interpretation page contains typical-weather cautions alongside older model and financial assumptions. Do not merge those assumptions into current V8 capability or quote its numerical accuracy bands as a guarantee for your house. No actual home model run or predicted output is created here.
Match the forecast to one identified proposal or installed configuration
Keep the complete proposal or as-built equipment schedule and the exact forecast version together. Identify the module quantity and specifications, inverter arrangement, roof sections and any later replacement or expansion. Ask the provider which configuration the run represents. A forecast from the original sale may not describe equipment installed after a change, even when the marketing summary continues to show the same total panel count.
Distinguish nameplate power from energy over a period. The model documentation identifies system capacity in kilowatts and energy outputs over monthly or annual periods; those are different quantities. Ask whether the presented result is AC system energy, DC output or a sunlight quantity. Do not add unlike units or compare panel nameplate capacity with an annual bill total as if they were the same measure.
Request location and weather provenance separately
Ask for the run’s input location and the actual weather source used. The V8 response fields distinguish requested inputs from station or weather-file information. A named city printed on a sales page is not enough to establish those details. Ask how the chosen data represent this site and what limitations remain, especially when the roof is in an unusual local setting or a comparison uses a different dataset.
Typical-weather modelling is not a prediction of next year’s exact cloud, smoke or snow conditions. Retain the provider’s explanation of how weather differences affect the intended comparison. If an existing-system review concerns a particular completed operating period, ask whether a typical-year forecast answers that question or whether another qualified performance analysis is needed. Do not treat weather provenance as a household fault clearance.
Review roof planes and orientation against the actual drawings
Identify each array section’s capacity, orientation, tilt and mounting arrangement from qualified design records. Ask how a proposal with different roof planes was represented and combined. A single generic orientation can conceal a material difference between the model and a multi-roof design. Do not climb onto the roof or measure electrical components to reconstruct missing inputs; request the installer’s records and the relevant professional review.
Keep generic module and array classifications distinct from the actual product schedule. The V8 documentation includes several classifications and mounting or tracking choices. A selected category does not independently certify compatibility, roof suitability or the exact manufacturer’s performance. Ask why the choices fit this proposal and whether later changes require another run. No universal orientation, tilt or equipment recommendation is supplied here.
Use an input-audit table to locate explanations that are still missing
A practical review should identify which assumptions are documented and which merely use defaults. The table below is an original comparison framework, not an API request or a set of design values. Ask the provider for the actual evidence before accepting a material difference between two forecasts as proof that one installer is superior.
| Run item | Evidence to request | Do not infer |
|---|---|---|
| Model version and date | Dated complete run or export | Every estimate used the same model |
| Weather source | Dataset or file and relevant response information | A sales-page city proves the input |
| Array and inverter representation | Matched design and assumptions | Panel count answers every capacity question |
| Roof planes | Actual orientation and capacity by section | One generic roof fits the home |
| Loss allowances | Reason and scope of each assumption | A default is a measured site loss |
| Outputs and warnings | Complete results with units and limits | An annual headline proves a guarantee |
Ask how losses and inverter effects are handled without double counting
The official system-loss help distinguishes its general allowance from inverter conversion efficiency and temperature effects calculated separately. Ask the provider which inputs account for the actual shading, soiling, availability or other stated losses and why those assumptions are appropriate. Do not manually add every displayed category to another output or subtract a second allowance for a process already represented by the model.
V8 also includes inputs beyond the simplest headline fields, including monthly irradiance-loss and bifacial-related options. Their presence is not proof that your run used them or that the selected values were measured onsite. Ask for the actual setting and its basis when it materially affects the forecast. This guide supplies no default-loss recommendation, degradation rate, snow-removal procedure or claim that one modelling choice recovers guaranteed energy.
Keep a forecast audit separate from diagnosing production changes
A well-documented forecast can become context for an operating review, but a gap from it does not identify a failed component. Establish whether the monitoring series is current and complete and whether the provider’s comparison accounts for the relevant weather and actual configuration. A partial month or an export trace may not match the forecasted production quantity. Ask for the proper qualified analysis rather than interpreting a single shortfall as a replacement instruction.
Our seasonal-production guide concerns an operating decline, while the clipping guide concerns a positive power plateau and design-review decision. Those are separate questions from the provenance of a forecast. Leave roof access, electrical measurements, protective settings and hardware changes to qualified personnel; a revised model run does not authorize physical intervention.
Compare alternatives with matched questions and stated differences
Ask providers to explain material input differences before comparing their annual headlines. A higher result might reflect another roof layout, capacity, weather file or loss assumption rather than a better outcome for the same installation. Request a documented comparison that keeps unchanged factors identified and explains each deliberate alternative. Do not select a system from a forecast ranking that omits the actual scope or relevant constraints.
If a financial claim accompanies the energy estimate, ask for its separate assumptions: the actual current tariff, household use, export treatment and contract context. The older PVWatts caution page’s financial language is not used here as a present utility rule. This article calculates no bill savings, payback, incentive or return. An energy estimate and an accepted current financial or contractual analysis are distinct deliverables.
Retain a baseline that another reviewer can reproduce
Keep the run date, provider, model version, design reference, capacity and roof-plane inputs, weather provenance, loss assumptions, complete results, units and warnings. Ask the provider to correct factual mismatches and explain the limitation of each accepted assumption. Preserve the earlier run too, so a changed forecast can be traced rather than silently replaced by an unrelated PDF headline.
Note which future changes would require re-evaluation, such as a revised array layout or replaced equipment. Keep the forecast with the solar ownership and maintenance file, but label it as modelled evidence. Do not describe it as a field measurement or a warranty. Any contractual performance promise needs its own actual document and appropriate review.
Illustrative scenario: the panel capacity matches but the estimates differ
Imagine two proposals list the same panel capacity while one forecast uses a different roof orientation and an unexplained loss allowance. This is an illustrative scenario, not modelling we performed. The owner asks for dated runs, matched roof-section records and explanations of the differences before drawing a conclusion. No numerical output, price or winning proposal is invented.
The checklist highlights known inputs, defaults, unresolved site assumptions and the provider’s corrected comparison. It also keeps typical-weather production apart from next year’s actual operation and from a tariff-based financial claim. The useful result is a transparent forecast that answers an identified question, not a promise that actual monthly production will follow the table or that the current system needs a larger inverter.
Frequently asked questions about PVWatts estimates
Is a default loss a measurement from my roof? No. Is an annual estimate a prediction of the next exact year? No. Is a modelled result a field test? No. Do kilowatts and kilowatt-hours mean the same thing? No. Does a higher headline establish a better provider? Compare the complete inputs and scope first.
Can an older help page supply universal accuracy or financial guarantees? No. Do API documentation examples describe my home? No; no API call or account was created for this guide. Does a forecast mismatch authorize changing electrical settings? No. What should I ask for first? The complete dated run, matching design, actual assumptions and limitations, so the estimate can be understood rather than merely repeated.