The most defensible comparison uses the same area and climate with verified whole-assembly U-values; bills still include ventilation, hot water, gains, behavior and system losses.
1. Define exactly what surface changes
Choose one wall, roof or floor boundary and measure its net area consistently. Exclude openings if their U-values are not changing, or model them as separate elements.
Use whole-assembly U-values. A material label alone omits linings, structure, air films, repeated studs, fixings and other thermal bridges.
2. Match degree-days to the intended indoor baseline
Heating degree-days compress a year of outdoor temperatures into one climate input. Record the source, location, years and base temperature because two published values may use different baselines.
The simple degree-day method is most useful for comparing the same building assumption before and after. It does not reproduce hourly solar, wind, occupancy or thermostat schedules.
3. Separate delivered heat from purchased energy
Fabric improvement reduces heat delivered to the space. The fuel or electricity reduction also depends on seasonal system performance, distribution losses and controls.
For a heat pump, use a seasonal performance factor rather than a mild-weather catalogue COP. For combustion, use an appropriate seasonal delivered-heat efficiency on the same energy basis as the tariff.
4. Expect the bill to contain other loads
Space-heating bills can include hot water, pumps, fans, standing losses and fixed charges. Air leakage and ventilation heat loss also remain unless the project changes them.
Comfort take-back is real: occupants may choose warmer rooms after insulation. That can be a successful comfort outcome even when the bill falls less than the modelled maximum.
5. Compare normalized evidence after the work
Record meter readings, degree-days, setpoints and occupancy before and after. Compare similar periods and normalize for weather rather than judging from one warm or cold month.
Investigate gaps between estimate and measurement: incomplete area, changed ventilation, thermal bypasses, moisture, control settings and system performance are more useful leads than hiding the difference.
Keep each layer of the estimate visible
| Layer | Main input | What it predicts |
|---|---|---|
| Fabric | Area and U-value change | Heat-flow reduction per degree |
| Climate | Heating degree-days and base | Seasonal temperature exposure |
| System | Seasonal performance | Purchased-energy reduction |
| Tariff | Marginal energy price | Indicative variable-cost saving |
| Verification | Meters, weather and use | Observed whole-building change |
Worked example
Compare one wall upgrade
A 120 m² element improves from U = 1.20 to 0.30 W/m²·K. The climate input is 3,000 K·day/year, system performance is 0.90 and energy costs 0.15 per kWh.
Heat saved = 120 × (1.20 − 0.30) × 3,000 × 24 / 1,000 = 7,776 kWh/year Purchased energy saved = 7,776 / 0.90 = 8,640 kWh/year Indicative cost saving = 8,640 × 0.15 = 1,296 per year
Treat 1,296 as a transparent scenario, not a guaranteed bill reduction. Test other U-values, climate periods, system performance and prices before comparing project costs.
What to have ready
- Net area of the element being changed
- Existing and proposed whole-assembly U-values
- Local degree-days with stated base and period
- Seasonal heating-system performance
- Comparable marginal energy price and meter history
Common mistakes
- Using insulation-product R-value as the whole wall.
- Mixing degree-day values with different base temperatures.
- Treating heat saved as electricity saved for every heating system.
- Calling a scenario result a guaranteed financial payback.
Reference check: 5 September 2026. Sources include US public agencies. Principles are general; regulations and design values must be checked where you live.Worked examples use stated hypothetical inputs.