How much does a thermostat setback save?

How does this calculator work?

Two published anchors, one scaling rule, and no hidden constants. The formula below is the formula the calculator executes; there is no second model behind it.

Step 1, scale the anchors to your schedule: scale = (depth ÷ 8.5) × (hours ÷ 8), capped at 2.2. DOE’s figure is quoted at a 7-10°F setback (midpoint 8.5°F) for 8 hours a day, so a schedule at exactly those conditions scales by 1.0.

Step 2, the ceiling: ceiling % = 10 × scale, applied to heating spend and to cooling spend alike. This is DOE’s "as much as 10%".

Step 3, the floor: heating floor % = 4 × scale and cooling floor % = 5 × scale, from the ENERGY STAR weighted 20th-percentile certification criteria. Heating and cooling differ because the criteria differ.

Step 4, the heat pump exception: If the system is a heat pump on an ordinary thermostat, the heating floor is not computed at all and the output says so. There is no fudge factor in that branch, because a fudge factor would be a number we made up.

HyreHVAC analysis: Step 1 is the only part of this that is ours, and it is the part to be skeptical of. DOE published a point, not a curve.

Linear scaling is a reasonable reading near the stated conditions and an increasingly poor one away from them, which is why the scaling is capped at 2.2× rather than allowed to run to an absurd number on a 20°F, 18-hour input.

What does a worked example look like?

A household spends $1,000 a year heating and $600 a year cooling. They hold 68°F and want to drop to 60°F while out, eight hours a day, five days a week.

They enter 8 hours because the schedule is what it is on the days it runs, and the calculator does not model weekends separately. The system is a gas furnace with central air.

Scale: (8 ÷ 8.5) × (8 ÷ 8) = 0.94.

Ceiling: 10% × 0.94 = 9.4%. On $1,000 of heating that is $94; on $600 of cooling, $56.

Floor: 4% × 0.94 = 3.8% of heating, or $38; 5% × 0.94 = 4.7% of cooling, or $28.

Result: $66 to $151 a year: Costing nothing, on a thermostat they already own.

Now change one thing. The same household has a heat pump instead, on the plain thermostat that came with it. The cooling side is unchanged at $28 to $56, because a cooling setback has no auxiliary-heat analogue.

The heating side becomes "up to $94, with no defensible floor", and the combined figure becomes "up to $151 a year, floor not quantifiable". Nothing about the house changed. What changed is how much anyone can honestly claim to know.

HyreHVAC analysis: That second result is the whole reason this page exists.

A thermostat marketed on "save up to 10%" is quoting the top of a band whose bottom, for one household in seven, is not a published quantity.

The saving is still probably positive. It is the confidence that is missing, not necessarily the money.

What does this model leave out?

Thermal mass: A heavy masonry house coasts down slowly and recovers slowly; a light-frame house does both quickly. The same schedule produces different results in the two, and neither DOE’s figure nor ours carries a term for it. In a heavy building a short setback saves noticeably less than linear scaling implies.

The recovery period itself: We treat the setback hours as saving and the rest of the day as unchanged.

In reality the recovery consumes part of the saving even on a furnace, because the system runs harder for a while.

The published anchors already have this baked in at their own stated conditions; away from those conditions, our scaling does not track it.

Rate structure: A flat cents-per-unit assumption. Time-of-use rates, demand charges, tiered residential rates and winter-peak pricing all change the answer. A setback timed to end during an on-peak window can be worth materially less than one that ends before it.

Occupancy and behavior: Every figure assumes the schedule is kept. It is not modeled that a household member overrides it four evenings a week.

The size of the auxiliary-heat penalty: Named repeatedly on this page and quantified nowhere on it, because no federal source quantifies it. This is a gap in the model that we are pointing at rather than papering over.

Anything about the equipment: This tool takes your system as it is. Whether it should be replaced, resized or serviced are different questions with much larger numbers attached, and they are answered elsewhere on this site rather than smuggled into a thermostat page.

HyreHVAC does not install, service or sell HVAC equipment, and holds no data on quotes, invoices or finished jobs. Nothing here is a price we observed. That is also why these tools can tell you to keep the system you have.

Who is most exposed to the heat pump recovery problem?

The setback advice everyone repeats was written for a house with a burner in it.

A furnace or a boiler burns nothing while the house coasts down, and it recovers on the same burner at the same efficiency it always had. Nothing about the schedule changes what an hour of heat costs.

That is the case the 10% figure describes, and for that case the advice is simply correct.

A heat pump is a different machine.

Its cost per unit of heat depends on how hard it is being asked to work and how cold it is outside, and when it cannot climb fast enough on its own the control calls auxiliary heat — which on the overwhelming majority of installed systems is electric resistance.

Resistance heat is the most expensive heat in an American house. An hour of it can undo several hours of coasting.

Source fact: DOE’s Building America Solution Center explains it: adaptive-recovery models are "equipped with an adaptive-recovery technology that directs the heat pump to minimize the use of costly auxiliary heating, which is typically electric resistance heat, when returning the space from a set-back temperature to the desired temperature."

HyreHVAC analysis: Read that sentence backwards and it is a description of the failure mode. If the thermostat’s job is to minimize auxiliary heat during recovery from a setback, then a thermostat that is not doing that job is permitting it.

The Residential Energy Consumption Survey puts 13.9% of US households on a heat pump for heating and 10.3% on a smart thermostat of any kind. Those two populations do not have to overlap, and there is no reason to assume they do.

The households where the standard advice is most likely to misfire are, in aggregate, the ones least likely to be running the control that stops it misfiring.

Why is the floor a certification threshold, not a study?

Almost every thermostat savings figure in circulation traces back to the same DOE guideline, and that guideline is a ceiling: as much as 10%. A ceiling on its own is not a range.

To give the reader a floor we needed a number that came from houses rather than from a rule of thumb, and there is exactly one federal figure that qualifies.

Source fact: To carry the ENERGY STAR label, a smart thermostat’s service provider must "use EPA-provided software to analyze and combine a year of data from hundreds of their customers' homes, reflecting how the thermostats were actually used, to calculate national savings metrics for heating and for cooling."

The thresholds are a lower 95% confidence limit of the weighted national average of at least 8% heating and 10% cooling runtime reduction, and a weighted national average of 20th percentiles of at least 4% heating and 5% cooling.

EPA glosses the second of those itself: "The 20th percentile means that 4 out of 5 homes in the sample saved at least that much."

HyreHVAC analysis: That 20th-percentile line is the most useful sentence in the whole subject, because it is a statement about the unlucky end of the distribution rather than the average. It is what makes an honest floor possible.

We use 4% and 5% as the bottom of the band at DOE’s stated conditions and scale them exactly as we scale the ceiling.

Note what this floor is not: it is a threshold a product must clear to be certified, not a measurement of your house, and it describes whole-thermostat behavior across a year rather than the effect of one scheduled setback.

The useful next step: If you own a smart thermostat, it almost certainly reports your own runtime by month.

That is a far better input to this question than any national figure, including both of ours — and it is the only way to find out which end of the 20th-to-95th-percentile spread your own house sits at.

Do you save a bigger percentage in a mild climate? Yes

The instinct is that a cold place has more to gain from a setback, because there is more heating to save. The percentage runs the other way.

Source fact: DOE’s Building America Solution Center states it plainly: "The percentage of savings from thermostat setback is greater for homes or buildings in mild climates than for those in more severe climates."

HyreHVAC analysis: The mechanism is that a setback saves in proportion to the reduction in the temperature difference across the building envelope, and that difference is what drives heat loss.

Drop the indoor temperature 8°F when it is 60°F outside and you have cut the driving difference by most of what it was. Drop it 8°F when it is 5°F outside and you have barely dented it.

The dollar saving may still be larger in the cold place, because there is more energy to save a slice of. But the fraction is smaller, and it is the fraction that every published guideline is quoted in.

This is also where the two complications compound rather than cancel. Deep winter is when the percentage saving is smallest and when a heat pump is most likely to need auxiliary heat to recover. The night the setback looks most tempting is the night it is worth least and risks most.

The useful next step: Look up your county’s heating and cooling degree days before you decide how much of your bill this is worth chasing.

A county with a heavy HDD figure and a light CDD figure is telling you that the heating side dominates your spend and that the percentage on it will be at the modest end.

Where should you go next?

Degree day lookup gives your county’s heating and cooling degree days, which is the direct measure of the climate severity DOE’s mild-climate qualifier turns on. Look at the ratio between HDD and CDD before deciding which half of this calculator matters to you.

Heat pump vs furnace running cost is the arithmetic underneath the system-type selector here, in far more detail — including where the balance point sits and what auxiliary heat costs when it runs.

SEER2 savings calculator answers the equipment version of this question. It is a much bigger number and a much bigger bill, and it shares this page’s discipline about not confusing a rating with a bill.

The maintenance calendar covers the things a homeowner can actually do that change running cost, in the order they are worth doing. A schedule is on that list. So are several things that outrank it.

Cold climate heat pump performance is the underlying study on how heat pump output falls as it gets colder — which is the physical reason the recovery problem exists at all.