Personally this doesnt make sense to me. if two things are of similar cost but one is more precise, you wouldnt even use or buy the less precise one. what makes sense is to buy a cheaper variant even IF it is less precise as long as it is precise enough for xx percent of targets. there is no situation in my mind where imprecision is welcome. you can always choose to hit near a target on purpose if that is absolutely needed.
I don't think he's suggesting there isn't a meaningful cost difference between the two versions, but rather that it should be framed as "one version at X cost was able to exceed the requirements in terms of precision" and "one version at a cost lower than X was able to meet the requirements in terms of precision".
The two variants shouldn't be compared with each other, but instead with the requirements they were intended to meet.
Ultimately all resources are finite, so if one variant was able to greatly exceed the requirements but at the cost of... a higher cost, and if exceeding that capability was not seen as essential or useful, then it could make quite a bit of sense to develop a lower cost version that was better calibrated to meet the requirements which also happened to use less resources... it means that one spends less resources on procuring the same target no. of total missiles (regardless of composition of the two variants), or they can expend the same amount of resources but procure a greater no. of total missiles than initially intended.