Methodology

How the Efficacy Index is calculated

Every figure in the index is reproducible from the source datasheet and the steps on this page. If you cannot reproduce a number, that is a defect and it will be fixed.

Compiled by Mark Riley, Cutter Electronics · 2026 edition · Back to the index

Derived, not measured. The 2026 edition contains no bench measurements. It applies engineering derates to published manufacturer figures. A future edition may include measured data, and it will be labelled separately if so.

The calculation

Claimed efficacy is the datasheet lumen figure divided by the wattage it is quoted against. Delivered efficacy applies only the loss factors that the stated claim basis has not already accounted for:

delivered = claimed x optical x driver x thermal chip basis apply optical, driver and thermal module basis apply driver and thermal system basis apply thermal only

Applying a driver derate to a system-level figure would count the same loss twice. This is the most common error in efficacy comparisons and the index avoids it by recording the claim basis for every product.

Factors and defaults

Where a datasheet states a value, that value is used. Where it does not, the default below is used and the row is marked ⓘ in the table.

Factor	Default	Basis for the default
Driver efficiency	0.88	Constant-current LED drivers in commercial luminaires typically test between 85% and 92%.
Optical efficiency	0.9	Luminaire optical efficiency (light output ratio) for a high bay reflector or lens.
Lumen maintenance at 50,000h	0.85	Used only where no LM-80 or TM-21 projection is published.

Thermal derate

Datasheets quote lumens at a rated ambient, almost always 25°C. Luminaires do not run at 25°C, and LED output falls as junction temperature rises. The index applies a derate from the rated ambient to a realistic in-service ambient.

thermal = 1 - (0.0035 x (service_Ta - rated_Ta))

The service ambient is set per category. This matters: a high bay sits near roof level in an uninsulated shed and runs during the day, while a street light runs outdoors at night. Applying a single figure to both would overstate the street light derate badly and make the two categories non-comparable.

Category	Service Ta	Why
`high-bay`	45°C	Mounted near roof level. Daytime operation in an uninsulated shed.
`downlight`	45°C	Recessed into a ceiling void with restricted airflow. Higher again where IC-rated and in contact with insulation.
`panel`	35°C	Recessed into a conditioned office ceiling with a plenum above.
`batten`	35°C	Conditioned interior. Surface mounting runs cooler than recessed.
`street`	25°C	Outdoors, operating at night. Daytime ambient is not relevant to when the luminaire runs.
`area-flood`	25°C	Outdoors, operating at night.

Any category not listed uses 35°C.

The coefficient of 0.0035 per °C reflects typical mid-power LED behaviour in a well heatsinked fitting. A poorly heatsinked fitting will do worse. These are deliberately conservative figures: the index is not trying to make products look bad.

Maintained efficacy

The 50,000 hour column is delivered efficacy multiplied by the lumen maintenance factor. It answers a different question from the delivered column: not what you get on day one, but what the installation still produces years in. Lighting designs that only satisfy a lux target on day one fail their own brief later.

When a product is excluded

A product is left out of the index, rather than estimated, in any of these cases:

No lumen figure, or no wattage the lumen figure is quoted against.
A lumen range spanning a whole family of LED counts or drive currents, with no single configuration identifiable.
Two conflicting power figures on the same datasheet. Where a sheet tabulates one input power but states another elsewhere, and the two differ materially, there is no defensible pair to publish. Picking either one would misrepresent the product.
No retrievable manufacturer source. Retailer and marketplace listings are not used.

Exclusions are applied on the same terms to every manufacturer. If a product you expect to see is missing, one of the above is why, and asking will get you a straight answer.

What this index does not do

It does not measure products. It does not rank brands on quality, build, warranty or support. It does not account for optical distribution, glare or colour rendering, all of which matter and none of which reduce to a single number. A high delivered efficacy figure does not by itself make a fitting right for a space.

Source data

Every row links to the manufacturer datasheet it was derived from, with the date it was retrieved. Datasheets change without notice. If a link is dead or a figure has been revised, report it and the row will be updated.

Corrections and revision history

Corrections are published, not silently applied. Manufacturer-supplied system-level test data always takes precedence over a derived figure. Report a correction