Our own numbers, and exactly how we get them.
The Bench Lab is the measured half of this site. Every figure we publish under a protocol below comes from a test we ran ourselves, on a sample we bought ourselves, using the method written on the protocol page. Specs quoted from a manufacturer are labelled as such everywhere they appear, and are never mixed into a bench result.
51 measured results across 51 products and 7 protocols. Slots we have opened but not yet measured are shown as pending rather than filled with an estimate. The whole dataset is downloadable from the data hub.
Edit Score rankings
Every reusable in the catalogue ranked on our own published criteria — durability 25%, everyday performance 25%, sustainability 20%, value 15%, materials 10%, evidence strength 5% — with each position and pillar available as a citable claim.
Bench cards
A bench card sets one product's manufacturer specification beside our own measurement of the same property, so it is obvious which figures have been verified and which are still the maker's word.
Carry load hold
kg held 60sThe load a bag holds for 60 seconds without seam, handle or fabric failure.
Cost per year over five years
currency/yearPurchase price spread across a conservative five-year service life, so products of different prices compare directly.
Dishwasher cycles to visible failure
cyclesHow many standard dishwasher cycles a product survives before finish, seal or print degrades.
Heat retention over six hours
°C at 6hHow hot a filled vessel still is six hours after sealing, measured at room temperature.
Lid teardown and reassembly
secondsHow long it takes to fully disassemble a lid for cleaning and put it back together correctly.
Measured empty weight
gThe weight of the product as it is carried: fully assembled, dry and empty, including the lid and any sleeve supplied in the box.
Seal integrity (inversion and shake)
ml escapedHow much liquid escapes a sealed vessel when it is inverted and shaken like a bag commute.
What we will not claim
We do not publish a number we did not measure. We do not convert a manufacturer's "keeps hot 12 hours" into a bench result. We do not average across samples we did not buy. Where a protocol produces a failure rather than a figure — a lid that cannot be taken apart by hand, a bag that tears at the first step — we publish the failure.
Sample sizes are small and stated on every row. A three-sample heat retention run tells you the difference between four hours and six; it does not tell you a manufacturing defect rate. Read the tolerance line on each protocol before quoting a difference of a degree or a millilitre.