Design for an ELV, not an RLV. The marginal cost benefits from repeated flights will still exist, but without the issues imposed by an RLV. Design a clustered core like the Atlas V, Delta IV, and Falcon 9, but size the core so a 1 core launch is about 20 tons to orbit total mass in LEO, crew comes out of that). For a three-core launch, that gets you...80 or so? Two of those beats a Saturn, and can lift as much as the most basic SDHLV. Design an big capsule for LEO. Maybe make it able to do BEO as well, I don't know. 4 or 6 person capsule, something like that, plus a SM. Maybe design a pressure fed LRB for the core to match 1xCore flight capacity to missions.
So now, if you fly a few medium satellites or science payloads a year (1x core launches), plus one or two orbital flights (1x core launches), plus a large station component (think a space station built out of a few Skylab-sized modules) or BEO flight (3x core flights) a year, you're talking about 7 launches a year, with maybe 10 cores ordered. That's the sort of area on the cost per launch chart where things take a serious turn for the "not too much" area. And I think without the orbiter reprocessing, that's doable.
So the idea is a common-core ELV starting in the medium range and able to be in the heavy range. Fly it often so economy of scale works for you, and without having a crew on every single flight and a touchy orbiter, there's fewer constraints on usage. Spend the extra money on getting the station flying earlier and going back BEO. Do that and you'll end up with a capable vehicle that will exceed expectations, not merely meet expectations continually revised downwards.