How Potrero calculates your solar and battery estimate

The estimate begins with the address, the bill range you select, and public or licensed property data available for that address. The address determines the roof, PG&E climate territory, local electric taxes, permitting jurisdiction, and any location-based project costs. When available, home size helps improve the no-bill usage estimate.

The important limitation is what we do not have: your utility’s interval meter history or a completed site survey. The instant estimate therefore knows the likely annual amount of electricity you use and the typical timing for a similar local home, but it does not claim to know exactly when your appliances ran last year.

We use the Google Maps Platform Solar API for building-specific roof geometry, sunlight, panel layouts, and annual DC energy estimates. We rescale that production to Potrero’s proposed panel wattage and apply a system-loss allowance for the inverter, wiring, soiling, and other real-world losses.

Google’s maximum panel count is a theoretical layout. For the instant estimate, we currently treat 70% of that theoretical maximum as the installable ceiling to leave room for fire setbacks, access paths, obstructions, and details that satellite imagery cannot confirm. The final engineered layout replaces that allowance with an actual design.

The Google result supplies the roof-specific annual production. We use NREL PVWatts v8 for the local month-by-hour production pattern and as a conservative fallback when Google does not return a usable production curve.

If you select a bill range, that range becomes the primary signal for estimated annual electricity use. If you do not, we start with PG&E’s published climate-zone baseline allowance, scale it to estimated average use, and adjust it for home size when reliable size data is available. The PG&E E-1 tariff supplies the baseline allowances; the home-size anchor comes from the EIA Residential Energy Consumption Survey.

An annual total is not enough to value solar and a battery because electricity prices and export credits change by hour. We distribute the annual use across the year with county-level NREL End-Use Load Profiles for detached single-family homes. NREL built these ResStock profiles from physics-based home models calibrated with data from 11 utilities, 2.3 million meters, and 15 end-use metering datasets.

Without your interval meter data, filling 8,760 rows would add detail without adding household-specific information. The production model instead uses 24 hourly periods for a representative day in each of the 12 months. That 288-period structure preserves the seasonal and time-of-day information supported by our inputs and makes it fast enough to compare many system sizes while you adjust the estimate.

We tested that choice against 300 individual NREL-modeled homes across Potrero’s ten PG&E territories. Each benchmark home was simulated chronologically for all 8,760 hours and compared with the production method across 15 solar, usage, and battery scenarios. The production model’s profile-weighted median year-one savings bias was +0.9%. The typical middle-50% household range still spanned about 7% of the median estimate, which is why we disclose household timing as uncertainty instead of pretending the estimate is a meter-specific forecast.

That study also calibrated the model’s direct solar-use adjustment. We count 88.5% of the direct solar-and-load overlap produced by the representative profiles; the remainder is moved back to grid imports and exports to account for clouds and appliance cycles that monthly averages smooth out. Battery-routed energy is not given an additional haircut because the chronological benchmark did not support one.

For each representative month, the model follows the battery’s state of charge through every hour until it reaches a repeatable daily cycle. Excess solar charges the battery. The battery then serves the home when doing so avoids the most expensive imports, and it can export during the limited late-summer windows where published export credits justify it. Grid charging is not included.

Normal bill-saving dispatch can use 80% of nameplate capacity. The other 20% is held as the standard on-grid reserve. If a homeowner requests additional outage reserve, that is held above the standard 20% and reduces the energy available for bill savings. The estimate assumes 90% system round-trip efficiency, below the standalone ratings published for the EG4 WallMount battery and FlexBOSS21 inverter, so component losses are not ignored.

The bill-savings model does not predict how long the home will run during an outage. Backup duration depends on the loads selected for backup, the battery state of charge when the outage begins, solar production during the outage, and the system’s commissioned reserve settings.

The before-project bill is an estimate on PG&E Schedule E-TOU-C, not a claim that this is the homeowner’s current tariff. The after-project bill uses Schedule E-ELEC, which PG&E uses for residential customers on its Solar Billing Plan.

The model prices imports by season and time of day, and prices exports with PG&E’s published Solar Billing Plan values. It also includes the monthly Base Services Charge, non-bypassable charges that exports cannot offset, the California Climate Credit on both bills, and applicable city or county utility-user taxes. PG&E explains the customer-facing rules and links its current export values on the Solar Billing Plan page.

The instant estimate assumes bundled PG&E generation before and after solar. Community Choice Aggregation providers can have different generation charges or credits, so CCA customers should treat the result as a PG&E-bundled comparison until their actual bill is reviewed.

The estimate prices every feasible combination of panel count and one to three battery modules that fits the roof and the homeowner’s selected features. A system expected to pay back within 15 years outranks one outside that guardrail. Within the same guardrail class, the model recommends the system with the highest estimated 25-year net present value, with payback and upfront cost used to break close ties.

Project cost comes from Potrero’s current installed pricing, location adjustments, selected add-ons, and address-qualified rebates. The estimate shows the Tax Credit Program price or the cash price according to the option selected by the homeowner.

This is a planning estimate, not a production guarantee, utility-bill guarantee, or final construction quote. The largest unknown is the home’s actual timing of electricity use. Weather, new loads, occupant behavior, future tariffs, CCA generation charges, shading that imagery misses, final equipment settings, and site conditions can all change the result in either direction.

The next level of accuracy comes from reviewing actual utility interval data and completing the site survey and engineered design. Those steps replace population-level assumptions with information from the specific home.