I got a tip of a tutor of mine that is really into recording pop music. He uses auto tune in real time when recording. SO he is monitoring the vocals trough the device. That way he already knows which words/syllables will work and which won't.
Then another tip for all this pitch correction stuff. If you need to correct by huge amounts (= half steps or more) it is best to cut out all the noisy phonemes and place them on a separate track. Nosiy phenomes are esse for example (or use 2 channels nd use the side chain monitoring feature of a deesser to isolate the esses on one track). That is because noisy stuff tends to be treated wrong with pitch correct software. Longer tonal phoebes like vocals. Another trick is to automate the correction amount on the vocals. then you can cut the correction on noisy phonemes and increase it on only the syllables you need to correct.
I know it is a lot of work what i told you, but when you just apply the auto tune you won't get commercial like results.
So my workflow is-> But auto tune on with very little correction -> the automate the correction amount so that it cuts out on noisy stuff-> increase the amount on really off tonal phonemes