Because I've prepared properly, the first take is often the one I end up editing and sending to my client. The second is simply there as a safety net, in case I need to replace any mistakes while editing. It's much better to capture such drop‑ins during the same session as the main performance, as you can be sure that the mic will be in the same place and your voice will have a similar character in both takes.
At this point — unless you're tracking for broadcast applications (where they don't work to peak levels) — consider 'normalising' your recording to an optimal level. Normalisation can be used to push the loudest peak to around ‑1dB and increase the volume of the entire recording by the same ratio. Of course, you can also tighten up dynamic range (the difference between the softest and loudest parts of your recording) by manually reducing the loudest sections of the recording before you apply normalisation. On a final note, if you do plan to use a compressor, then you might as well add gain at that stage instead of normalising.
Not all noise can be tackled in this way, though: you need to listen for clicks, plosives, digital glitches and the like. These can normally be acceptably repaired by using a 'heal' tool, or a pencil tool to redraw the waveform. Popped 'p's can often be 'fixed' using a high‑pass filter set at 100Hz. For a single glitch, you can zoom in and cut out the cycle of the waveform in which the glitch appears. Just be careful to start and end the cut where the waveform crosses the centre line, otherwise you'll inadvertently add another digital glitch. If glitches are frequent, it's likely that there's a problem with your audio interface's buffer settings — it may be just a playback issue.