admin July 18th, 2008
I am reading a fascinating book now on the subject of randomness and how it affects our lives. What has that to do with wine marketing you may ask? Well, according to the author, it actually has a lot to do with wine ratings, which as we all know, continue to play a very influential part in wine marketing. Indeed, when I was at the UC Davis wine executive confence in March, most of the participants thought that ratings were more important than ever to sales.
The book by Leonard Mlodinow is The Drunkard’s Walk: How Randomness Rules Our Lives. There’s a very interesting discussion of wine ratings in the book. Now the whole subject of ratings is obviously controversial (unless your wine just got a 95 in which case it is fabulous) but even if you think “expert” ratings are valuable, the book raises an interesting statistical objection which is how little real value a single critic’s rating has without the context of additional ratings from either the same critic or from others. Basically, the objection is that a single rating based on subjective assessment is simply a sample point on a continuum of assessment. Any single rating of the wine (even if it was by Mr. Parker) is subject to a significant potential for random error. It is only when you get one critic or multiple critics to repeatedly rate the same wine that you get any kind of meaningful assessment.
This point is an interesting one because I have always been fascinated by: 1) the power of ratings, particularly individual ones, and 2) the frequency with which ratings vary between critics and sometimes with the same critic.
For example, if you have a subscription to erobertparker.com, you can go there now and check out his ratings for Chateau Latour a Pomerol from the 2000 vintage. He has rated that wine on 3 separate occasions, giving it 87-88 the first time, then 90-92 and finally 91. There is a 5 point spread there from the highest to lowest rating. More importantly, the spread goes between a 92 rating (where the wine would fly off the shelves) to an 87 (where it would likely languish). Even more interesting is to throw other critics into the mix. If you look on the Wine Spectator web site, there is a single rating of Latour a Pomerol 2000 from James Suckling … but that is 95 points. So now you have an 8 point spread. The average of these ratings, incidentally, is 91 so the owners of the chateau would probably be happy in any event. However, this does seem to bear out the point that a single rating is not very accurate or at least should be qualified by a variation: in this case +/- 4 points. Somehow I don’t think Robert Parker or any critic (and certainly not retail stores) are going to say this wine is a 91 +/- 4 points but technically that would seem to be a better and more meaningful rating.
For this example, I also picked a relatively well known wine from a good vintage. The variation on ratings seems much higher for lesser known wines. For example, I recently tried a bottle of Ojai Bien Nacido Chardonnay from the 2005 vintage (which I liked). Robert Parker gives that wine 91 points saying it is “outstanding” and that it exhibits “elegant lemon rind, orange zest, and citrus notes”. Over at the Wine Spectator, they must have had a different wine, because James Laube gives it only 82 points, describing it with a “trim beam of spice, citrus and green apple”.
There’s a subscription wine rating service out of San Francisco that averages ratings and ranks by value, winebluebook.com, but they don’t provide any indication of the variations in the ratings. However, in reality, consumers are unlikely to understand or act on any of this.
If a wine gets 90 or above in any single rating, it is likely that retailers will continue to post the highest score. However, if your wine gets relegated to the basement of 80s ratings (or worse), you can at least take comfort in the laws of randomness … which mean that your wine will likely return to its statistical ratings norm the next time around, even if it honestly tasted no better or worse on this occasion.
