Regeneration is the reverse trade: a strong sodium chloride brine drives the calcium and magnesium back off the resin and down the drain, and sodium takes their places. The controller meters that brine by volume, but the setting is written as a mass per volume of resin.

$m_{\text{salt}} = D\,V$ — m-salt equals D V. $m_{\text{salt}}$ is the salt drawn per regeneration in pounds, $D$ is the salt dosage in pounds per cubic foot of resin — the number programmed into the controller — and $V$ is the resin volume in cubic feet. Solve it forward for the salt bill, or backwards for $D$ when a brine-tank log and a known bed size are all you have.

Now the trade-off, because this is where the money is. At 6 lb/ft³ a standard resin returns roughly 20,000 grains per cubic foot — about 3300 grains per pound of salt. At 15 lb/ft³ it returns roughly 30,000 — about 2000 grains per pound. More salt buys more capacity and worse salt efficiency, every time. What it also buys is lower leakage, and on a boiler or a laundry that is the reason the high dosage gets chosen anyway.

Which brings us to leakage. $R = \dfrac{C_{\text{in}} - C_{\text{out}}}{C_{\text{in}}}$ — R equals C-in minus C-out, over C-in. $R$ is the removal efficiency, a fraction or a percent; $C_{\text{in}}$ is the inlet hardness and $C_{\text{out}}$ the outlet hardness, both in mg/L as CaCO₃ and both in the same currency or the ratio means nothing. Rearranged, the leakage itself is $C_{\text{out}} = C_{\text{in}}(1 - R)$.

Carry this one out of the room: a percentage is not a specification. A softener advertised at 99 % removal sounds faultless, and on 300 mg/L raw water it leaks 3 mg/L — three times over a boiler maker's typical 1.0 mg/L feedwater limit. Percentages flatter hard water and punish nobody. The boiler reads milligrams per litre, so compare the outlet against the specification, never the percentage against a feeling.