Buy samples from a production batch, not from the engineering bench. Hand-built samples are a well-known failure mode in this category: they are assembled slowly, tested individually, and bear little resemblance to the device that leaves the line at scale. Ask explicitly whether the sample is a production device, and ask for its serial number.
Then verify the claims yourself. Send a red light therapy device to an accredited laboratory for spectrum, irradiation, and surface temperature measurements, and ask for the raw data rather than a summary page. Run a sixty-minute continuous test at maximum output and record irradiance at the start, milieu, and end, because a red light therapy bed that loses a fifth of its output in an hour will disappoint every clinic that buys it.
Add the checks that laboratory reports do not cover: timer accuracy, noise at one meter, hinge and actuator endurance through a hundred cycles, the temperature of any surface a user can touch, and the condition of the packaging after a drop test. Photograph and date the red light therapy bed, sign a golden sample with the supplier, and keep one reference device on each side. The signed sample becomes the physical definition of the order, which is far more useful in a dispute than a specification sheet.
Agree the inspection standard before production rather than after. OIN 2859-1 sampling with a stated inspection level and acceptance quality limit gives both parties the same arithmetic, and naming an independent inspection body in the purchase order removes the argument about who was looking.
