Our methodology
Every figure on this site is an estimate produced from public data. This page explains exactly how, so you can judge how much weight to give it.
Where our data comes from
We build estimates from four kinds of public source, cited with a consultation date on every guide:
- U.S. Department of Energy (Energy Saver) — system efficiencies, savings guidance, best practice.
- Energy Star — certified product performance data and efficiency criteria.
- U.S. Energy Information Administration — residential energy prices (Electric Power Monthly) and household consumption patterns (RECS).
- NREL — solar production modelling via PVWatts.
Retail prices come from current listings at major retailers and are stamped with the date we checked them.
How our calculators work
Each calculator shows a “How we calculate this” block containing its actual formula, its assumptions, its limits, and its sources. Nothing is hidden behind a black box, and none of them ask for an email to show a result.
Results are deliberately rounded to the nearest $10 and prefixed with ≈. A figure like “$2,637.41” would imply a precision we do not have. Every result carries a note that it is an estimate for guidance, not a quote.
The assumptions we make
Unless a guide says otherwise, our baseline is:
- A 2,000 sq ft single-family home in a mixed U.S. climate zone.
- 2026 national-average residential energy prices.
- No federal tax credit applied — the 25C and 25D residential credits terminated for property placed in service after 31 December 2025. State and utility rebates are excluded too, as they vary too much to generalise.
- Equipment operating at its rated efficiency, correctly installed and maintained.
Each guide restates the assumptions that matter for its own numbers, in an “assumptions and limitations” block.
The TrimWatt Savings Score
Method version 1.1 — 22 July 2026.
It is a judgement expressed as a number. We chose the components, the weights, and the band boundaries, and a reasonable person could choose differently. What it is useful for is comparing actions against each other on a consistent basis, and stopping one flattering figure — usually annual savings — from hiding a long payback or a licensed-trade install. It scores an action, never a product, and we never score a product as tested.
The formula
Total = payback (0–40) + materiality (0–15) + effort (0–20) + comfort (0–10) + confidence (0–15)
payback years = upfront cost ÷ annual savings
| Component | Weight | Bands |
|---|---|---|
| Payback speed | 40 | ≤1 yr: 40 · ≤3: 34 · ≤5: 28 · ≤8: 20 · ≤12: 12 · ≤20: 6 · beyond: 2 · no saving: 0 |
| Materiality of the saving | 15 | ≥$500: 15 · ≥$250: 12 · ≥$120: 9 · ≥$60: 6 · ≥$20: 3 · below: 1 |
| Effort to do | 20 | Level 1: 20 · 2: 16 · 3: 12 · 4: 6 · 5: 2 |
| Comfort impact | 10 | +2: 10 · +1: 8 · 0: 6 · −1: 3 · −2: 0 |
| Data confidence | 15 | High: 15 · Medium: 9 · Low: 4 |
Bands: 80+ Strong, 60–79 Good, 40–59 Situational, under 40 Weak.
The overlap between payback and materiality — and why we kept it
These two components share an input: annual savings appears in both. That is a real weakness and we are not going to hide it. We kept a scaled-down materiality component because payback is a ratio and is blind to scale: a $15 gadget saving $15 a year pays back in twelve months but is not worth anyone's attention, while a $9,000 system saving $800 a year matters even though it pays back slowly.
To limit the double count, materiality carries only 15 points rather than the 25 it had in method v1.0, and the freed weight moved to two components that share no input with payback: effort (15 → 20) and data confidence (10 → 15). Savings-derived components therefore account for 55 of 100 points rather than 65. If you want a pure return-on-money view, read the payback line on its own and ignore the total.
Scoring criteria — effort
| Level | Test |
|---|---|
| 1 | No purchase and no tools — a setting or behaviour change only. |
| 2 | Consumer product, hand tools only, no connection to mains electricity, gas, or plumbing; under an hour. |
| 3 | DIY involving low-voltage (24V) wiring or reversible work; up to half a day. |
| 4 | Needs a trade for part of the job, or touches 120/240V circuits, ductwork, or roof access. |
| 5 | Licensed trade required by code (gas, refrigerant, service panel); permit and/or inspection likely. |
Scoring criteria — comfort impact
| Level | Test |
|---|---|
| +2 | Directly removes a documented comfort problem (cold rooms, draughts, humidity) in most homes. |
| +1 | Small improvement, or added convenience with no offsetting downside. |
| 0 | No perceptible change in temperature, noise, humidity, or convenience. |
| −1 | One minor trade-off most households tolerate (e.g. cooler while asleep, audible equipment). |
| −2 | A trade-off many households notice daily and may reverse. |
Scoring criteria — data confidence
| Level | Test |
|---|---|
| High | Figures come from a primary federal source (DOE, EIA, Energy Star, NREL) and the result is dominated by arithmetic, so the spread across typical U.S. homes is narrow. |
| Medium | Figures come from a primary federal source, but the result depends on climate, local rate, or existing equipment, so real outcomes vary widely. |
| Low | No primary federal figure exists for this specific outcome, or the result depends mostly on individual behaviour and site conditions. |
When we do not score
If we have no defensible cost or savings figure, the component prints “Not enough reliable data to score” instead of a number. We would rather show a gap than manufacture a score.
Limits and the risk of over-simplification
- It compresses a decision into one number. That is the point and the danger — a total can look authoritative when the underlying spread is wide. Read the components.
- The weights are editorial. We think payback deserves the largest share; if you rent, or plan to move, effort and comfort should probably outweigh it for you.
- Bands are step functions. A payback of 5.0 and 5.1 years score differently. Treat totals within a few points of each other as equivalent.
- It uses national baselines. A high score is not a promise for your house; your rate, climate, and existing equipment move it.
- It ignores non-energy value entirely — resale, noise, air quality, and the fact that failing equipment needs replacing regardless.
- It is not a product rating and carries no testing claim.
Charts and tables
Every chart on this site is generated from the figures cited in that same guide — we do not reproduce third-party visuals. Bars are drawn from plain arithmetic you can repeat, and each chart states the assumptions and the rate used underneath it. Where a chart shows costs by electricity rate, the range reflects the real spread across U.S. states rather than a single national average.
What we do not do
- We do not run a test lab and do not physically test products. Our reviews say Researched or Compared, never Tested, unless we genuinely tested the item.
- We do not model your specific home. A home energy monitor or a professional audit will always beat any calculator.
- We do not present estimates as guarantees, and we do not publish savings figures we cannot trace to a source or an assumption.
Keeping figures current
Energy prices, incentives, and product prices move. Every guide displays a “figures last verified”date, and every roundup shows when it was last checked. If the date looks stale, treat the numbers with caution and tell us — see the editorial policy for how corrections work.