You mixed a batch and it came out wrong. How much more salt, or how much RO/DI, to land it on target.
The batch is 3.00 ppt short. Add the salt in stages with the powerhead running, and remeasure between. Salt dissolves faster than a refractometer reading settles, so give it twenty minutes before believing the number.
salt = 94.6 L × 3.00 ppt × 1.0334 = 293 g
Adding salt to an existing batch is the difference calculation, not the full one. You need salt for the gap between where the batch is and where you want it, which is V × (S₁ − S₀) × f.
Coming down is conservation of salt. The salt already dissolved is fixed, so adding fresh water spreads it thinner: RO to add = V × (S₀/S₁ − 1).
When the mixing vessel is already full, remove V × (1 − S₁/S₀) litres and replace it with RO/DI instead. Same result, same total volume.
Add roughly three quarters of the calculated salt, circulate, then measure again and add the rest. Undershooting costs a few more minutes. Overshooting costs a bucket of RO water and an hour.
Give a refractometer reading time to settle. Salt dissolves in minutes but the batch is not homogeneous until it has circulated, and reading too early is the usual reason a batch seems to need correcting twice.
If a batch is more than a few ppt off what you expected from the salt you weighed, the instrument is the more likely culprit. Calibrate with 35 ppt standard fluid, not with RO water, because a zero point calibration does not catch scale errors at the top of the range.
Sources: Red Sea Salt manual (PDF) · Fritz RPM Reef Pro Mix · UNESCO EOS-80 / Millero & Poisson 1981