Every LED manufacturer publishes a savings figure, and almost all of them are built on assumptions that do not describe your building. The arithmetic itself is simple — it is one multiplication — but the inputs are where the estimates go wrong. This article works through both, and then does the sums for two real buildings.
The formula, and why it is the easy part
The annual running cost of any lamp is:
Annual cost
Cost = (Watts × Hours per day × 365 ÷ 1000) × Rate per kWh
The middle section converts watt-hours to kilowatt-hours, which is the unit your utility bills in.
For a single 60 W incandescent, on four hours a day, at 18 cents:
60 × 4 × 365 ÷ 1000 = 87.6 kWh × $0.18 = $15.77 a year
Its 10 W LED replacement, all else equal:
10 × 4 × 365 ÷ 1000 = 14.6 kWh × $0.18 = $2.63 a year
A saving of $13.14 a year from one fitting. The multiplication is trivial. What is not trivial is knowing that the lamp is 60 W, that it runs four hours a day, and that you pay 18 cents.
The four errors that ruin most estimates
1. Guessing burn hours
This is the largest error by a wide margin, and it is almost always an underestimate. People report the hours they remember using a light, not the hours it was on. In the audits we run, hallway and porch lamps come back at two to four times their reported burn time, because nobody counts the evening they left the landing light on or the porch light that is on a photocell.
The fix is to measure. A plug-in energy monitor on a lamp circuit for two weeks, or a circuit-level logger for a whole building, converts a guess into data. Where logging is impractical, use these defaults, which are conservative:
| Location | Typical hours/day | Notes |
|---|---|---|
| Living room | 4–5 | Strongly seasonal; double in winter |
| Kitchen | 3–4 | Plus under-cabinet task lighting |
| Bedroom | 1–2 | Often overestimated |
| Hallway / landing | 5–8 | Almost always underestimated |
| Bathroom | 1–2 | Many short cycles |
| Exterior / porch | 8–12 | Photocell or dusk-to-dawn |
| Office (commercial) | 10–12 | Often 24/7 in circulation areas |
| Retail floor | 12–16 | Plus display lighting after hours |
2. Using an average tariff
National average electricity prices are a statistic, not a price. Residential rates in the US vary by more than a factor of three between states, and within a state a time-of-use tariff can vary by a factor of two between afternoon peak and overnight. Lighting load is concentrated in the evening, which on many time-of-use tariffs is the most expensive band — so using an average rate systematically understates lighting cost and therefore understates the saving.
Take the rate from your bill. If you are on time-of-use, take the rate for the band your lights actually run in.
3. Ignoring replacement labour
In a house, changing a bulb is free. In a warehouse with 9-metre ceilings it requires a scissor lift, two people and a permit. Commercial retrofit business cases that count only energy routinely miss half the saving, because maintenance access is the dominant cost for high-bay and exterior lighting. Count the truck rolls.
4. Forgetting the cooling interaction
Every watt of lighting in an air-conditioned space becomes a watt of heat the air conditioning has to remove. At a typical coefficient of performance of 3, removing that heat costs another third of a watt. So in a cooled building, lighting savings are roughly 1.3 times the direct electrical saving during the cooling season.
In a heated building in winter the reverse applies — the old lamps were contributing heat. This effect is real but much smaller than it sounds, because lighting heat is delivered at the ceiling where it is least useful, and because heating is often gas while lighting is always electricity, so the substitution is between different fuels at different prices.
Worked example: a four-bedroom house
A real audit, anonymised. Forty-two fittings, mixed technology, tariff 19.4 cents per kWh flat rate. Burn hours from fourteen days of circuit logging.
| Location | Lamps | Before | After | h/day | kWh saved/yr | $ saved/yr |
|---|---|---|---|---|---|---|
| Living room | 6 | 60 W halogen | 10 W LED | 5.2 | 570 | $110.58 |
| Kitchen spots | 10 | 50 W halogen | 4.5 W LED | 3.8 | 631 | $122.41 |
| Hall & landing | 5 | 60 W incand. | 10 W LED | 7.1 | 648 | $125.71 |
| Bedrooms | 8 | 40 W incand. | 5 W LED | 1.6 | 164 | $31.81 |
| Bathrooms | 6 | 35 W halogen | 4.5 W LED | 1.9 | 127 | $24.64 |
| Exterior | 4 | 60 W incand. | 9 W LED | 10.4 | 775 | $150.35 |
| Utility / garage | 3 | 100 W incand. | 17 W LED | 2.2 | 200 | $38.80 |
| Total | 42 | — | — | — | 3,115 | $604.30 |
Lamp cost for the retrofit: $214. Simple payback: 4.2 months. Over the ten-year life of the lamps, net saving after lamp cost is about $5,830.
Two things stand out. The exterior lighting — four lamps out of forty-two — produced a quarter of the saving, because it runs for ten hours a day. And the bedrooms, where the owner expected the biggest win because there are eight of them, produced 5%. Burn hours dominate everything.
Worked example: a small office
2,400 square feet, 64 fittings, mostly 2×36 W fluorescent troffers with magnetic ballasts. Tariff: 14.1 cents energy plus a $17.40 per kW monthly demand charge. Occupancy 11 hours a day, 5 days a week, 52 weeks: 2,860 hours a year.
- Existing load: 64 fittings × 2 × 36 W = 4,608 W, plus ballast losses of about 12% = 5,161 W
- Existing energy: 5,161 × 2,860 ÷ 1000 = 14,760 kWh/yr → $2,081
- LED panel replacement: 64 × 32 W = 2,048 W
- New energy: 2,048 × 2,860 ÷ 1000 = 5,857 kWh/yr → $826
- Energy saving: $1,255 a year
- Demand saving: 3.11 kW × $17.40 × 12 = $649 a year
- Cooling saving: roughly 30% of the energy saving over a 6-month cooling season → about $188
- Maintenance saving: 64 tubes relamped every 3 years at $9 fitted = $192/yr, now effectively zero
Total annual saving: $2,284. Installed project cost: $6,150, less a $1,280 utility rebate = $4,870 net. Simple payback: 2.1 years.
Note the demand charge
The demand charge contributed 28% of the saving and appears in almost no manufacturer’s calculator. Lighting runs during the working day, which for most commercial buildings is when the monthly peak is set. If your bill has a kW line on it, that line is part of the business case.
Payback, and when it is the wrong question
Simple payback — net cost divided by annual saving — is the standard measure and it is fine for a domestic retrofit, where the numbers are small and the horizon is short.
For larger projects it is a poor decision tool, for three reasons. It ignores everything after the payback date, so a project paying back in three years and running for fifteen looks identical to one paying back in three years and running for four. It ignores the cost of capital. And it ignores risk — specifically, the risk that energy prices move.
For anything over about $10,000, calculate net present value over the equipment life at your own discount rate, and run the energy price up and down 20% to see how sensitive the answer is. Lighting retrofits are unusual in that they generally look better under sensitivity analysis, because the saving is denominated in energy and energy prices have historically risen.
When the answer is “not yet”
Sometimes the honest answer is to wait. A lamp burning under an hour a day in a guest room saves about a dollar a year; replacing it is a rounding error. Fittings due to be removed in a planned refurbishment should not be retrofitted twice. And a space whose use is about to change should be relit for the new use, not the old one. We put these in the “do not do” column of every audit report, because a report with nothing in that column is a sales document.
Rebates, incentives and where to find them
Most US utilities run commercial lighting rebate programmes, and many run residential ones. They typically pay per fitting or per watt reduced, and they frequently require pre-approval before work starts — which is the single most common way money is left on the table.
- Prescriptive rebates pay a fixed amount per qualifying product from a published list. Simple, fast, and the product must be on the list — usually DLC or ENERGY STAR qualified.
- Custom rebates pay based on calculated savings for projects that do not fit the prescriptive list. More paperwork, usually more money, and they need a measurement and verification plan agreed in advance.
- Midstream or instant discounts are applied at the distributor, so the price you are quoted is already net. Ask whether a quote is gross or net of incentive; comparing one of each is how procurement decisions go wrong.
Federal and state tax treatment changes often enough that it is not worth quoting specifics here. What is stable is the advice: check before you buy, get pre-approval in writing, and keep the product spec sheets — rebate administrators ask for them months later.
Doing it yourself, in six steps
- Inventory. Walk the building and record every fitting: location, lamp type, wattage, quantity, base type, whether it is dimmed, and its condition. A spreadsheet is enough.
- Measure burn hours. Two weeks of logging on representative circuits, or a data logger on the ones you are least sure about. Do not skip this step; it is where the accuracy lives.
- Get your real tariff. From the bill: energy rate, any time-of-use bands, any demand charge, and the standing charge.
- Calculate the baseline. Fitting by fitting, using the formula above. Sum it. Compare against your total bill as a sanity check — lighting is typically 8–15% of a modern home’s electricity and 15–25% of a commercial building’s.
- Specify replacements. Matched on lumens, not watts, with the colour temperature and CRI the space needs. Then recalculate.
- Rank by saving, not by cost. Do the exterior and circulation lighting first. They are boring, nobody notices them, and they are where the money is.
Or put your numbers into our savings calculator, which runs the same arithmetic, and then have us log the building if the answer looks big enough to act on.
Common questions
In a home that has already switched to LED, typically 5–8%. In a home still on incandescent or halogen, 15–20%. In commercial buildings, 15–25% before retrofit and under 10% after. If you want the real figure rather than a range, a circuit-level logger will tell you in a fortnight.
No, and it never was for LEDs. That advice came from fluorescent tubes, whose starters wore on each strike. LEDs have no start cost and no switching penalty. Turn them off whenever you leave the room.
Yes — typically 0.2 to 0.5 W to keep the radio listening. Across forty smart lamps that is 10–20 W continuously, or roughly 130 kWh a year. It is not a reason to avoid them, but it should be in the calculation.
