[G0HZU_JMR]

Yes, it's hard to say much without knowing more info. But I'd guess the coupling factor will be around 10dB as I think it has to couple out about +5dBm and the LO level at the coupler is probably in the region of +15dBm. I also think a lot depends on the type of coupler and how much effort went into getting good performance.

But I can offer a few ballpark numbers for a typical PCB based 10dB coupler made from coupled transmission lines. Assuming the coupler has been optimised for through VSWR and a 10dB coupling factor then (without a termination at port P3) I'd expect the input return loss at port P1 in the image below to degrade to approx double the coupling factor. So the input return loss would degrade to about 20dB. This is because I'd expect the input reflection coefficient to degrade to 0.1 in this case. This is a fairly insignificant change. This happens because of the reflected signal (at port P3) that is coupled back through the coupler to port P1. These numbers assume the coupler was optimised for input match during the design phase. So the optimised return loss would degrade because of the degradation in the reflection coefficient at port P1 when P3 was left unterminated.

When used in an ideal 50R system I think any change in mainline loss would be tiny/negligible with a typical 10dB coupler. I also think the match at port P2 should be virtually unaffected when port 3 is left unterminated. So the overall impact would be negligible on a typical printed coupler. But this coupler is presumably a wideband design as it has to work over the full range of the first LO and it may be that the coupling factor won't be that consistent over its full range. But it may still hold true that the input return loss could degrade to twice the coupling factor.
Of course, in the real system the coupler won't be in a perfect 50R environment so it will be harder to predict what will really happen without a a termination at port P3. But I doubt it will be anything significant?

But the Tek coupler could be completely different because we don't know much about it? Not all couplers will behave like the example I quoted above, but quite a few will, I think.