Every trap in a catalogue looks convincing. The useful question is how far its signal travels. A UV lamp produces a pool of light a few metres wide; a CO₂ plume drifts tens of metres and dilutes as it goes. That is the whole argument for host-mimicking traps, and it explains why two units with the same wattage can differ by an order of magnitude in catch.
Carbon dioxide is not a poison and not an insecticide. It is a cue. A resting adult human releases roughly 100–200 ml per minute, in a plume that bends with the wind and pools in sheltered air. A trap that releases gas in bursts, or at the wrong height, produces something that does not resemble a host — and the catch reflects it.
This is also why CO₂ traps are usually specified with a regulator rather than a fixed orifice. Sites differ: a walled garden holds a plume; a windy terrace strips it. Adjusting flow is cheaper than adding another unit.
What CO₂ does not do is select species. It will bring biting midges and some flies as well. That is normal for a host-mimicking trap, and it is worth saying plainly in a listing rather than letting a customer discover it.
Our own rule of thumb: decide the siting first, the gas flow second, and the hardware third. A modest trap in the right corner outperforms a large one in the wrong one.