MARKERS

The first notable thing is that you can look for the markers (guaiacol and 4-methylguaiacol) in the grapes.  If you find it at all, I would recommend not even wondering about threshold levels.  This is because the extracted levels in a wine are always much higher than the measurable levels in grapes/juice and it is nearly impossible to predict where the wine will end up, or whether that level will be offensive. The safe thing to do is assume there is a problem and act accordingly at this point.

MACHINE HARVESTING

Especially for whites, it is best to avoid machine harvesting if at all possible, if smoke taint is present.  This next thing is unproven: if you can't avoid machine harvesting, you may want to separate (especially with whites) the "free run" juice in the bottoms of the bins.  This will have been macerating for an indeterminate amount of time. After primary fermentation you can send out labs and recombine if it is not much higher than the other fractions. For all machine harvested fruit, make it as high a priority as you can to get it picked cold and get it processed as fast as you can. For reds, machine harvesting will make your taint levels slightly higher, but it probably won't make the different between "offensive" and "inoffensive."

WHITE FERMENTATION

For white wine, there are only a few things you can/should do at this stage.  Eliminate all MOG prior to juicing, it may have higher concentrations of volatile phenols.  Avoid intentional skin contact.  Don't whole cluster press.  Separate your press fractions as rigorously as you are able to. Don't depend on your palate.  At this stage some smoke taint is readily apparent, and some is not. Strangely, this seems to be independent of concentration.

RED FERMENTATION

For red wine, there are a few more things to do in addition to the above.  Perhaps the most important one may be avoiding the cold soak.  This is completely non-scientific and non-rigorous.  However, we have observed at least 3 large vineyards where fruit is sold to multiple wineries.  For each winery making the same variety, from the same vineyard, the g/4mg levels varied significantly.  We compared harvest dates and ruled that out as an influence (again, sample is not large enough to reach a definitive conclusion).  What did seem to correlate is the presence (or length) of a cold soak.  The higher levels had longer cold soaks and the lower levels had short or no cold soak.  These three vineyards, interestingly, are all Pinot Noir.

That said, another approach is to make the wine exactly as you would normally. If you are going to have to treat your wine anyway, treating it a little more may be worth it to make the wine you really want to make in every other aspect.

POST FERMENTATION

Most importantly, get the wine off of the lees as soon as you are able. They do seem to be able to adsorb some additional amounts of the smoke related compounds with extended lees contact.  Again, if the wine is clearly smoky and needing treatment, weigh the tradeoff here if lees contact is an important part of your style objectives.