HeatPumpAtlasUSA
Methodology: How Every Number on This Site Is Produced

Methodology

How these numbers are produced

Every figure on this site is either a published government statistic or a modelled estimate with a stated method. This page separates the two, lists the assumptions, and explains what the model cannot know.

Written by HeatPumpAtlasUSA Editorial Updated How we calculate this

Four rules

How we keep the numbers honest

01

Verified vs modelled

Electricity prices are verified. Installed cost is modelled. The site never presents a model output as a measurement.

02

Publish the losing cases

In many states a heat pump is the wrong financial answer. Saying so is the only way the tool stays trustworthy.

03

Assume the data rots

Prices move monthly. Programs move weekly. Every page has an update date; rebate pages also have a verification date.

04

Never invent a number

Where we cannot support a figure, the calculator says so and widens the range instead of filling the gap with an average.

What is verified data

These inputs are published statistics, refreshed on a schedule, and traceable to a primary source.

  • Electricity prices

    EIA Electric Power Monthly table 5.6.A, state residential average. June 2026 is the most recent published month at the time of this revision. EIA publishes roughly two months behind and revises prior months.

  • Natural gas prices

    EIA Natural Gas Monthly table 20, state residential per thousand cubic feet. May 2026 used for the current revision.

  • Heating oil and propane

    EIA weekly reference prices. These are regional and seasonal, and the national reference price is used where state data is thin.

  • Climate

    NOAA Climate Prediction Center population-weighted heating degree days, state monthly and seasonal accumulations.

  • Heating fuel mix

    US Census American Community Survey table B25040, share of occupied units by main heating fuel.

  • Housing vintage

    Census ACS table B25034, year structure built, used as an envelope efficiency input.

What is modelled, and how

Everything downstream of those inputs is a model. Here is the whole model.

Installed cost

Cost bands come from published 2026 contractor and consumer pricing, cross-checked against regional contractor quotes and against published national ranges from consumer sources. They are then adjusted by a regional labor factor and, in high-cost states, an additional factor. The bands are wide on purpose: two contractors bidding the same house frequently differ by 40 percent, so a narrow band would be false precision.

Heating load

If you provide your annual heating spend, the calculator backs the heating load out of it using your current fuel and its assumed efficiency. That is the high-confidence path. If you do not, it models the load from floor area, heating degree days, insulation quality and build vintage. The second path is marked as lower confidence in the result.

Operating cost

Each option is priced per million BTU of heat actually delivered, which accounts for both the fuel price and the efficiency of the equipment. Heat pumps are modelled at a seasonal coefficient of performance rather than a rated lab figure, and ducted systems take a duct loss penalty. Natural gas is converted using 10.27 therms per Mcf.

Totals and payback

The ten-year total is undiscounted and covers up-front cost after open rebates plus ten years of operating cost. Payback is simple, not discounted, and the calculator reports no payback when the operating cost difference is zero or negative rather than producing an infinite or misleading figure.

Every assumption, in one table

If you want to check our arithmetic, this is everything you need.

Model assumptions and their basis

  • Electricity price

    Value used
    State residential average
    Basis
    EIA Electric Power Monthly, table 5.6.A

  • Natural gas price

    Value used
    State residential average per Mcf
    Basis
    EIA Natural Gas Monthly, table 20

  • Heating oil

    Value used
    $3.62 per gallon
    Basis
    EIA heating oil reference price

  • Propane

    Value used
    $2.58 per gallon
    Basis
    EIA propane reference price

  • Climate

    Value used
    Population-weighted heating degree days
    Basis
    NOAA Climate Prediction Center

  • Heat pump efficiency, cold climate

    Value used
    Seasonal COP 2.9
    Basis
    Published seasonal performance for cold-climate equipment

  • Heat pump efficiency, standard

    Value used
    Seasonal COP 2.6
    Basis
    Published seasonal performance for standard equipment

  • Ductless efficiency

    Value used
    Seasonal COP 3
    Basis
    No duct loss, published seasonal performance

  • Duct loss, ducted

    Value used
    15%
    Basis
    Published duct loss estimates for unconditioned ducts

  • Gas furnace efficiency

    Value used
    90% AFUE
    Basis
    Federal minimum and typical installed equipment

  • Oil and propane efficiency

    Value used
    82% AFUE
    Basis
    Typical installed equipment age mix

  • Gas heat content

    Value used
    10.27 therms per Mcf
    Basis
    EIA residential heat content convention

  • Grid carbon intensity

    Value used
    0.85 lb CO2 per kWh
    Basis
    US grid average, eGRID class factor

  • Heating load coefficient

    Value used
    0.0025 kWh thermal per sqft per HDD
    Basis
    Calibrated against published US space heating consumption
Constants are versioned with the site data revision. If you think one is wrong, tell us and we will publish the correction.

Regional cost factors

The same equipment costs different amounts to install in different parts of the country. This is the adjustment applied.

Regional labor factors applied to national cost bands

  • Northeast

    Labor factor
    1.15x
    High-cost states
    Massachusetts, New York, Connecticut, Rhode Island, New Jersey add 1.12x

  • West

    Labor factor
    1.20x
    High-cost states
    California, Washington, Hawaii, Alaska add 1.12x

  • Midwest

    Labor factor
    0.95x
    High-cost states
    None

  • South

    Labor factor
    0.92x
    High-cost states
    District of Columbia adds 1.12x
Factors are modelled from regional labor cost indices and observed contractor pricing, not from a survey. They exist so the calculator does not quietly treat a Boston install and an Oklahoma install as identical.

How rebates are handled

This is the part of the site most likely to be wrong, and we treat it accordingly.

We do not scrape a national incentive database and republish it. Programs change status faster than any automated feed we evaluated, and a stale "available" label is worse than no label because it can lead someone into an ineligible installation.

Instead, we maintain a short list of programs, verify each one against the administrator's own page, and record three things: the status, the verification date and the specific conditions. The calculator includes only programs marked open. Programs that are reserved or paused are excluded from the total and labelled, not removed, so you can recognise the programs that other sites are still advertising.

There is one pattern worth knowing about. Several programs require a reservation or a rebate commitment letter before installation, and applying afterward can disqualify the project. If a page tells you the amount without telling you the order of operations, it is incomplete.

What the model cannot know

The honest list. Every item here can move a real project off the modelled result.

  • Your actual utility tariff

    We use EIA state averages, which are revenue divided by sales. They include fixed customer charges and are not the rate on your bill. Someone on a time-of-use rate, a heat pump rate, or a municipal utility can be materially off.

  • Your duct condition

    We assume a standard duct loss. Leaky, undersized or poorly routed ducts in unconditioned space can lose far more than that, and the ductless-versus-ducted decision often turns on this.

  • Your measured air leakage

    Insulation quality is a three-level user estimate, not a blower door result. A tight house and a leaky house with the same square footage and vintage can differ by 30 percent on load.

  • Your contractor

    A correctly sized system installed well performs close to the model. An oversized system that short-cycles, or one installed with a poor air seal, will not. This is the single largest source of real-world variance.

  • Future energy prices

    The model uses current prices. Any projection over ten years is a bet, and we would rather show you a current-price total than a forecast we cannot defend.

  • Comfort and noise

    Model quality, indoor humidity in winter, and outdoor unit noise are real reasons people are satisfied or dissatisfied. None of them appear in a cost model.

Corrections

If we are wrong, we want to know, and we will say so.

Energy and incentive data changes constantly, and we will get things wrong. If you spot an error, particularly a rebate amount, a program status or a rate that has moved, contact editorial with the page and the source you are looking at.

Substantive corrections are updated on the page and the last updated date changes. We do not quietly edit numbers, because the update date is part of how you should judge whether to trust the figure.

Privacy and data handling

The calculator runs entirely in your browser. It resolves your ZIP against a bundled table, computes against bundled energy data, and sends nothing. There is no analytics call inside the calculator and no account.

If you ask us to email the estimate, that is a separate action. The request carries your ZIP, contact details and the scenario you can see on the form. We use it to send you the breakdown and a short list of installers who cover that ZIP. You can ask us to delete a submitted request by emailing us.

Full source list

Verified

Green dot marks a primary government, program or utility source. Grey dot marks a secondary benchmark or editorial reference. Our full method, including how modelled figures are produced, is on the methodology page.