Weather-normalized savings: why before/after bill comparisons lie

Published July 2026 · SolarVerdict methodology series

“We installed X and our bills dropped 30%” is the most common form of energy-upgrade evidence and one of the least reliable. If the winter after the install was mild, the bills would have dropped anyway. If it was brutal, real savings can look like failure. Weather normalization is the standard fix, and it’s simple enough to do on your own bills.

The tool: degree days

A heating degree day (HDD) measures how much colder a day was than a base temperature (conventionally 65°F): a day averaging 40°F contributes 25 HDD. Cooling degree days (CDD) mirror this for heat. Summed over a month, degree days are an excellent proxy for how hard your heating or cooling system had to work. Historical degree-day data for your location is freely available from public weather sources.

The method: a two-line regression

Home energy use decomposes cleanly:

Fit this line to a year of pre-upgrade bills (utility portals export monthly usage; pair each month with its HDD). The slope β captures your envelope and equipment; the intercept is everything else. To normalize: multiply β by a normal year’s degree days — a long-run average — instead of any particular year’s. To measure a real upgrade effect: fit the line again on post-upgrade bills and compare the two lines, not the two winters’ bills. A heat pump should show up as a lower β on the gas line and a higher (but cheaper per delivered unit) electric β; attic insulation shows up purely as lower β.

The solar side: typical years and real years

Solar has the same problem in the other direction. Production estimates come from typical meteorological year (TMY) datasets — statistical composites of long-run weather for your location. Any single real year deviates from typical: cloudier or sunnier by several percent is routine, and one bad year early in ownership can make a fine system look broken (or a lucky year make a mediocre quote look conservative). Honest treatment:

• Quote expected production as a range around the typical-year estimate, not a single number.
• Judge a real system’s performance against that year’s actual weather (measured irradiance), never against the brochure.
• Remember payback compounds weather with degradation: separating the two requires normalization first.

Why this matters for a purchase decision

Every savings claim you’ll ever be shown — installer quote, neighbor’s testimonial, online calculator — is implicitly a claim about weather. The neighbor’s heat pump “cut their bill in half” partly because last winter was gentle; the calculator’s payback assumes typical sun every year for 25 years. Normalization converts all of these to the same yardstick: a normal year at your address. That’s the yardstick a decision should be made on, and it’s the one every number in a SolarVerdict report uses — it’s the “weather-normalized” in the name.

Do it yourself checklist

1. Export 12–24 months of usage from your utility portal.
2. Pull monthly HDD/CDD for your nearest station (base 65°F).
3. Fit usage vs. degree days; note base load and slope.
4. Multiply the slope by normal-year degree days to get your weather-normalized heating/cooling consumption.
5. Evaluate any upgrade — solar, heat pump (break-even math here), insulation — against that normalized baseline.