There are areas with highly permiable soil and scattered watersheds. Much of the rain that falls in such places either isn't captured by reservoirs, or is directed elsewhere (absorbed into the ground, perhaps refreshing aquifers or the water table, evaporation). Reservoirs themselves have evaporative losses (especially in dry and windy climates). At smaller scales, localised cisterns could make sense, especially as supplemental storage and when integrated into initial designs (the case for many traditional building styles in dry climates).
Considering that existing building styles already provide virtually all that's necessary for rainwater capture except the storage containment itself, the additional marginal costs are fairly modest.
It’s possible to minimize evaporative losses by covering reservoirs. Cisterns don’t generally do that well by comparison because they have so much surface area exposed to dry air compared to their volume as well as leaks etc. https://www.awtti.com/reservoir-covers-info/
What’s not obvious is how cheaply these things contain water. The hover dam for example can hold back back over 14,000 m3 of water per 1 m3 of concrete. Ignoring profit from hydroelectric power in 2020 money it cost about 1$ to store ~13,000 gallons for 85 years and counting. With the added upside of minimizing flooding and hydroelectric power.
Of course that water is stored a long way from where it initially fell, but that’s a different story.
The cost vs. stored capacity question is an appropriate one. Looking around, rainwater capture systems vary in cost, though a rough average of about $2,500 for a 5,000 gallon system is given.
Keep in mind that the comparsion shouldn't be against the largest and most efficient reservoir (e.g., Hoover Dam), but against the best incremental option (marginal cost) of providing alternative impoundment.
I'd already pointed at part of that issue above: there are only so many Hoover Dam / Lake Mead opportunities available. Most have been exploited, and the environmental externalities are high. Rainwater capture is generally proposed where alternatives aren't viable (few homeowners can build a Hoover Dam on their own property, or even as a collective community project), where the terrain and hydrology don't support buillding dams, where environmental concerns are to great, or (and probably realistically the most significant issue) where legal claims on water rights make stream impoundments nonviable.
How this maps out in dollars per gallon over time remains unclear, though again, it's a good question.
(Lifetime is another key consideration. The anticipated lifetime of structures is another question, and Hoover Dam faces challenges on several fronts here, with both siltation and drought limiting effectiveness of the structure).
There are areas with highly permiable soil and scattered watersheds. Much of the rain that falls in such places either isn't captured by reservoirs, or is directed elsewhere (absorbed into the ground, perhaps refreshing aquifers or the water table, evaporation). Reservoirs themselves have evaporative losses (especially in dry and windy climates). At smaller scales, localised cisterns could make sense, especially as supplemental storage and when integrated into initial designs (the case for many traditional building styles in dry climates).
Considering that existing building styles already provide virtually all that's necessary for rainwater capture except the storage containment itself, the additional marginal costs are fairly modest.