As I’ve mentioned before, I love The Academic Wino. The site’s tagline is “Dissecting current research relating to wine.” But it could just as well be “We read these studies so you don’t have to.” The summaries on the site have often led me to seek out the original papers, and I’ve learned a lot that way.
Last week, The Academic Wino (I’ll just call it TAW from now on) summarized a market research study that looked at whether numerical ratings for wine affect sales. The full (and stuffy) title of the paper is “Expert Opinion and the Demand for Experience Goods: An Experimental Approach in the Retail Wine Market.”** As TAW reported, putting shelf tags with a score on wines in a supermarket increases wine sales for those wines over the untagged wines. Comparing a supermarket with the tags to one without showed that the shelf tags increased sales of the tagged wine compared to the store without tags.
The study also concluded that a higher score boosts sales, a lower score decreases sales, and that a score of 80 points seems to be the dividing line. The wine scores in the study ranged from 78 to 89.
Pretty impressive sounding, isn’t it? We’ve always wondered whether “Expert Opinion” matters for wine sales, and we’ve got anecdotal evidence to suggest it does. But no real data to back it up. TAW wasn’t too enthusiastic, though, because the study was extremely limited. I was intrigued, and looked at the original paper to see more details about how the study was done and what it concluded.
I had many of the same reservations as TAW did: the shelf tags were put on in only one supermarket (and comparing the data to 38 “control stores” without tags), the study lasted only one month, and the analysis didn’t specifically control for the price of the wine, which could cloud the results. But what do I know? I’m a chemical engineer and happy to critique scientific studies, but I’m not an economist. So I asked my sister Sue, who has worked in market research for over 20 years, to weigh in.
Sue confirmed that the study wasn’t thorough enough to merit more than a gee-that’s-interesting as far as its applicability to the real world of wine buying. Since the authors examined data at 39 supermarkets with high volumes of wine sales – one with shelf tags and 38 without – they missed an opportunity to make the study more significant by putting shelf tags in more of the supermarkets. Generally when companies do data tests in supermarkets they pick equal numbers of test and control stores, usually 10 of each. The stores are carefully matched on geo-demographics to minimize variations before testing and make for cleaner analysis. Sue also thought that one month wasn’t enough to get convincing results (although as a wine merchant, I can say that March, the month used for the study, is about as “normal” a wine-buying month as you can get.) She speculated, as did TAW, that either the authors couldn’t get permission to put tags in more than the one store or they didn’t have the resources to do it.
So, just like with studies on the positive health benefits of wine, we’re still waiting for something more definitive and applicable to the real world of wine buying and wine drinking. It seems at first thought that shelf tags with ratings would be a no-brainer since they appear to boost sales. But producing shelf tags (or buying them from the company that rated the wines) isn’t necessarily cheap, and assumes that a producer has done all the legwork getting his or her wine scored by the well-known experts first. Then someone has to put those tags up and make sure they stay with the correct product as it changes position on the shelves. If there are too many wines tagged with different sources for the scores, that might be confusing as well – is a “Wine Spectator” rating on one wine better than a “Wine Advocate” rating on another, for example – so it’s not clear that a producer (or the distributor) of a particular non-mass-market wine would decide that it’s worth the effort based on these data.
Looking at the study and talking with Sue reminded me that I’ve been curious for a while about just where data on wine sales come from. If you get the Wine Business Daily news e-mails, you’ll see links to plenty of articles discussing wine sales. Some of these articles reference data from industry-sponsored reports, and you can get the complete reports by paying for them, sometimes hundreds of dollars for each one. But if you’re cheap, like me, you’re reading about figures generated by A.C. Nielsen Company or Symphony IRI Group (IRI), the behemoths of sales data.
Back in early 2011, I did my end-of-year accounting and found that first vine’s 2010 sales were down from 2009. I asked a couple of other small, local wine retailers about their sales and found decreases of 10 to 20 percent. At the same time, though, news stories based on Nielsen data reported that wine sales in the U.S. were up seven percent in 2010 over 2009. The D.C. area always comes out on top for wine sales per capita, so I wondered how both sets of data could be true, and where Nielsen’s data came from.
We’ve all heard of Nielsen, thanks mostly to their television ratings work. Who among us hasn’t longed at one time or another to be part of a “Nielsen Family,” watching TV and having our viewing habits contribute to whether certain programs stay on the air or not? IRI doesn’t have the same public name recognition, but IRI and Nielsen together gather and analyze the vast majority of data on consumer packaged goods in the U.S., including wine. And many wine industry analyses rely primarily on Nielsen and IRI data too. Sue used to work for IRI, so I asked her about wine sales data and how they’re collected.
Now for the $64,000 question. Where exactly do these companies get their wine sales tracking data? The answer, surprisingly? Scanned checkout data from supermarkets and Wal-Mart, including Sam’s Club. And Total Wine, Centennial, Malloy’s Finest, Bottle King and some of the other big wine/liquor chains. That’s pretty much it.
No data from Costco, Whole Foods, Trader Joe’s, or H.E. Butt (a supermarket chain in the Southwest). And certainly no data from most small or independent wine/liquor stores, online wine sales sites, or winery to consumer sales.
Sue told me that Costco, Whole Foods, Trader Joe’s, and H.E. Butt all decline to cooperate with Nielsen and IRI, although they have scanned data. Certainly the two would love to have the data if they could get them. The two companies get some data from independent wine/liquor stores, if they scan and want to give or sell the data, and incorporate them into the national sales figures. Clearly Nielsen and IRI have a network of data from some independent wine/liquor shops (and some affiliated with supermarket chains, like Shop-Rite Liquors, for example) so they can project shop estimates in states like New York, which doesn’t allow supermarket wine sales. But it’s likely the companies aren’t hitting the majority of smaller or more specialized wine shops.
I’m not trying to be a snob about this – supermarkets sell a lot of wine for a reason. It’s nice to be able to pick up a bottle when you’re buying other things for dinner. And some supermarkets, like the Georgetown Safeway here in DC, have a good selection and reasonable prices, better than some local liquor stores.
But Costco is currently the nation’s largest wine retailer, and in urban areas like DC, Whole Foods and Trader Joe’s sell a lot of wine. Not to mention that they have a unique demographic that isn’t necessarily buying wine in more traditional supermarkets. And Costco has a sales model that changes wine stock more than twice a year, the way supermarkets typically do. While Nielsen and IRI may have developed some ways of taking a stab at their sales – and these companies occasionally release their own total sales data – they’re pretty much out of the loop for monthly or quarterly data. Each year I read comparisons of the all-important 4th quarter wine sales to the previous year’s 4thquarter, which probably leave Costco and the others out, so it’s not clear how accurate or relevant the data actually are.
Sue pointed out one other consequence of not having Costco in the data set. Costco sharply discounts wine, so the actual average sales price of a given wine that Costco also sells is probably lower than Nielsen or IRI would report. And it may also affect the amounts of wine sold nationally in different price categories.
Sue explained that Nielsen and IRI have a second data source, called panel data, which is used to gather more information about store scanner data. If the store data show what happens, the panel data can tell you why. Panel data also have the advantage of covering any and all outlets where consumers might purchase packaged goods, from supermarkets to online to social media to high-end wine shops. Nielsen and IRI have collaborated on recruiting “panelists” (balanced for demographics and geography) who shop wherever they like, and then use a scanner at home to record their purchases, along with where they bought the items and the prices. There are also opportunities to question the panelists about their purchases. So, for example, it’s possible to look at some people who haven’t bought wine for a few months but are now buying, or who switch from one label to another, and find out more about why they’ve done that. But the panels have always been a tool to understand more about a particular business, not a true substitute for weekly store scanner data from retailers, which track sales, share, distribution, etc.
Now that I have more information on wine sales data, I can put them in perspective. I don’t know what part of the wine universe Nielsen and IRI actually cover, but I have a better idea of what they don’t cover. So finally there’s at least a partial explanation of why first vine’s and other small retailers’ sales don’t necessarily match the national trends:
Once again, we were left out of the Nielsen family. And as your therapist will tell you, it always comes down to family in the end, doesn’t it?
Thankfully, most family interactions are trauma-free, like the trip to Alaska Cy and I took last month. My mom’s older sister moved there in the early 1950s, and married a man who was born in Alaska. I hadn’t seen them in decades. And although I’d met my cousins, I hadn’t met their spouses or children. It was a great visit, and nice to see that we all share a love of good food. Our last night there, my cousin Mel and I made a seafood chowder recipe that was invented by Mel’s daughter, Elizabeth. She enjoys experimenting with cooking, and this recipe is terrific.
You’ll notice that the seafood in the recipe is all in cans. You can get great canned local fish and seafood in Alaska. I also think of the cans as a reminder of the days when that’s all you could get there. My uncle was born and raised in a town called Naknek, on Bristol Bay in southwestern Alaska. Every single thing his family ate came in a can and a big load of canned food was delivered once a year. Milk came in a can, and so did whole chickens (I’ve seen canned whole chicken on “Chopped,” one of the Food Network’s chef competition shows, and it’s not pretty.) According to my uncle, Naknek had four salmon canneries in its heyday. Fishing boats came in with the salmon and it was processed immediately. From boat to can in no time. Using canned seafood makes the dish pretty much a snap because you can keep most of the ingredients around.
Some substitution is possible, depending on where you live. In spring and fall we can usually get fresh crab meat and possibly shucked fresh clams. If you can find them, go ahead and use them – make sure that the clams have some clam liquor with them and chop them into small pieces before adding them. You’ll want to rinse the crab meat in a little water (save the water to add to the chowder), pick through the crab for shell pieces and break the larger lumps up a bit.
But don’t be tempted to use fresh salmon in this recipe. Canned red Alaskan salmon is delicious and has just the right texture for the chowder. I found canned red Alaskan salmon at Whole Foods (which means, of course, that Nielsen and IRI can’t track the sales). If you can’t find it, try this website. You can also add a small can (3.75 oz) of smoked Alaskan salmon to the chowder for a unique, hearty flavor. Elizabeth discovered that browning the cubed potatoes helps keep them from disintegrating and adds nice flavor too, so don’t skip that step.
The chowder needs a wine with a little acidity to balance all that richness, so try Bodega Traslagares Verdejo ($13), which has great fruit flavors and a subtle floral note as well. A great combination of food and drink, perfect for any family gathering.
**Hilger, J., Rafert, G., and Villas-Boas, Sofia. The Review of Economics and Statistics, November 2011, 93(4): 1289-1296.
Serves 4 to 6
5 – strips bacon cut into ½-inch pieces (see note, below)
2 – medium baking potatoes, peeled and cubed
1 – yellow onion chopped
3 – cloves garlic chopped
2 – 8 ounce cans of minced clams, drained (reserve juice)
1 – 8 ounce can of crab, drained (reserve juice)
1 – 3 ounce can of smoked oysters, drained (reserve oil separately)
1 – 15 ounce can of red Alaskan salmon, drained, skin and bones removed, broken into small lumps (reserve juice)
2 – Tbs butter
3 – Tbs flour
½ – cup whole milk (approximate amount)
1 – cup half and half (approximated amount)
1 – Tbs chopped fresh rosemary
1 – Tbs chopped fresh thyme
½ – tsp salt
½ – tsp pepper
½ – tsp paprika
Fry bacon in large frying pan. When it is crisp add potato cubes and cook covered, stirring occasionally. Add onion and garlic to pan when potatoes are a little browned and almost done (still somewhat stiff to fork). Mix and cover, cook over medium heat until onions are softened (8 to 10 minutes). Add oysters and salmon and heat through. Put oyster oil and 2 Tbs. butter, over low heat, in a heavy stockpot. Add 3 Tbs. flour and mix with whisk over low heat for a minute. Stir in milk until the bottom of pot is covered and mix while it thickens. Add the seafood/potato mix and half and half just until things float. Stir and add the clam, crab, and salmon juices to the pot. (You can add additional milk/half and half to achieve desired thickness.) Cover and put on a low simmer for 45 minutes. About 20 minutes in add the herbs, salt, pepper, and paprika. After 45 minutes, taste for seasoning, then add the clams and crab and keep at a bare simmer for 10 minutes. Serve with lots of good bread.
Note: for those who want to leave the bacon out, you can brown some butter and use it to cook the potatoes and onions. Browned butter has great flavor. Melt 5 tablespoons of unsalted butter in a small saucepan over low heat. (Don’t use a non-stick saucepan, or you won’t be able to see the color change). Once the butter is melted, raise the heat a little and continue to cook. The butter will boil (because of the water in it) and then you’ll see the foam and solids on the top color and sink to the bottom. Swirl the pan and check carefully – when the liquid on top gets a little brown, immediately pour the liquid into a heatproof glass measuring cup, leaving the solids behind. Pour the lightly browned butter into the large frying pan, and immediately add the potatoes as directed in the recipe. If you’re not using bacon, you might want to add a small can (3.75 ounces) of smoked Alaskan salmon to the chowder, and use smoked paprika too.