Consumer body-fat scales use bioelectrical impedance analysis (BIA): they send a small, imperceptible current through your body and measure resistance. Fat resists the current more than lean tissue and water do, so the device estimates body composition from that resistance -- then runs it through a formula calibrated on population averages, not you specifically.
The problem is that resistance is driven heavily by hydration, and hydration swings throughout a normal day for reasons that have nothing to do with your actual body fat: a workout, a large meal, a glass of water, even the time of day. DEXA (dual-energy X-ray absorptiometry) uses a completely different method -- measuring how tissue absorbs two X-ray energy levels -- which makes it a stronger reference standard, though not a condition-proof one either.
Published comparisons of consumer BIA scales against DEXA commonly find gaps of several percentage points of body fat, and the size of that gap varies a lot by device, population and measurement conditions -- there isn't one universal correction factor, which is why this tool shows you a direction and a rough band rather than a single invented number.