Basically what I think you're saying is...
Fly the SCA back towards your track from the 'new' drift-corrected heading rather than the 'old' one that got you off track in the first place?
Why I doubt I ever really noticed the difference:
1) In your example 18 degrees is a huge track error! This translates to only a 1 mile difference.
2) Are you able to judge the 3nm off track so accurately? If it's further away than that your method actually makes things worse.
3) Small instrument errors, slight heading / speed inaccuracies, wind changes etc. all add up as well.
4) 90knots is a fairly sedentary TAS, anything faster and the effect becomes less and less pronounced.