How Much is Too Much?: A Study on Choice Overload in Wine Retail

So, you want to buy a bottle of wine. You walk into the wine shop, grocery store, or wherever wine is sold where you live and observe the selection. More often than not, you’re met with aisle upon aisle of hundreds

Photo by Flickr user  Geoffrey Fairchild (

Photo by Flickr user
Geoffrey Fairchild (

if not thousands of bottles. How does that make you feel? Are you overwhelmed? Exhilarated? Indifferent?

In retail, people can become overwhelmed by the selection to choose from, or what is otherwise known as “choice overload”. Studies in many different retail settings have shown that choice overload does exist and that it can lead to decreased customer satisfaction and decreased sales for a given store. In wine, however, the concept of choice overload has not been tested in a scientific setting until now.

Choice overload was first defined in the 1970s by Alvin Toffler, who stated that it occurs when “the advantages of diversity and individualization are canceled by the complexity of buyer’s decision-making process”. In other words, having a diverse selection of items confuses the consumer so much that they can no longer make a decision at all.

One example of choice overload in a retail setting was done in 2000 using jams in a supermarket. When consumers were given a choice of 6 different jams, 30% of them ended up purchasing at least one of them. When the choice was increased to 24 different jams, only 3% of the consumers purchased a jar, reducing the potential sales by 90%.

The idea of choice overload is further complicated when the consumer is no longer purchasing for themselves but instead is purchasing for someone else. In a very small study (only 68 participants) in two wine shops in NYC, customers who were purchasing for themselves reported greater satisfaction at the smaller wine shop while customers who were purchasing for someone else reported greater satisfaction at the larger wine shop. While this study did not look at choice overload explicitly, it does raise the question of whether or not it does exist in the wine retail setting.

As a result of the vast evidence in the retail setting in support of choice overload, a new working paper from the American Association of Wine Economists aimed to examine this phenomenon in the wine retail setting.

Several hypotheses were tested based on observations and analysis of other goods:

  • The more wines available for purchase in a store, the lower the customer satisfaction.
  • The more wines available for purchase in a store, the more difficult it will be for a customer to make a selection.
  • The more wine knowledge a customer has, the higher the threshold for choice overload (i.e. more knowledge = can tolerate many more wines for selection).
  • The more comfortable and satisfied a customer is with a particular store, the higher the threshold for choice overload (i.e. more devoted clientele will not be as phased by a larger selection than someone who shops around).


This study had three parts:

  • The first was using a self-reported questionnaire to determine consumer purchase behaviors and attitudes and how they know and understand the concept of choice overload.
  • The second was an observational study in a wine shop to determine how they actually behaved in a wine retail setting.
  • The third and final part was a live interview segment with some of those observed in the wine retail setting to determine how comparable their self-reported behavior was to their actual behavior in the shop.

The Questionnaire

The questionnaire utilized in this study asked many questions regarding wine purchase attitudes and

Photo by Flickr user Julie, Dave & Family (

Photo by Flickr user Julie, Dave & Family (

behaviors, including but not limited to how they browse and choose wine in a shop, if they ask for help or choose wine on their own, and how many bottles they typically purchase per visit.

Questions also asked were related to intent to purchase (i.e. did they come into the store knowing what they wanted), and how long they typically spent in the shop.

4000 surveys were sent out using email information from wine retailer’s databases at 9 different stores in the greater NYC area (NY, NJ, and CT). In order to get an accurate representation of the sample at a 95% confidence interval, they would need to receive at least 843 completed surveys back.

The Observations

For the interviews, three of the nine stores in the greater NYC area were chosen to try and get an accurate representation of the wine consumers in that area. Specifically, one store from each state of NY, NJ, and CT were selected.

Researchers used tracking and timing sheets to observe the behavior or each consumer through the wine shop. Additional information collected was the total time spent in the store, the time spent in each section of the store, and whether or not the consumer interacted with any staff at all.

Details regarding how the customer traveled through the store, how many and which bottles were touched, and how many/which bottles were ultimately placed in the cart for purchase were also noted.

Behavior based around a small subset of wine types were chosen for a more detailed analysis: Zinfandel, Cabernet Sauvignon-based wines, red Bordeaux, Pinot Grigio, and Argentinian wines.

Researchers made a note to attempt to observe at least 100 random customers.

The Interviews

Interviews were conducted after customers were seen to have made a purchase and were approached by researchers outside of the store.

Questions asked were related to if they knew what wine they wanted when they walked in, how satisfied they were with the wine they purchased, how satisfied they were with the store overall, their guess on how many wines they had to choose from before settling on the one(s) they purchased, and their perceived wine knowledge.


The Questionnaire

  • Out of 4,000 surveys emailed, 806 were received completed (note: 843 were needed to achieve a 95% confidence interval, so the sample size ended up being a little on the small side).
  • Most of the respondents self-reported as being high frequency wine consumers.
  • 57% of respondents reported purchasing wine bi-weekly or more.
  • 50% of respondents reported purchasing at least four bottles of wine at each visit.
  • 86% of respondents reported having plenty of time to browse the selection of wines at the shop.
  • 70% of respondents “totally disagreed” with the statement that having too many choices in the wine shop makes it more difficult to choose the wine they want, while only 5% “totally agreed”.
  • 83% of respondents self-reported as being happy with the selections at the wine shop.
  • 14% of respondents wanted a greater selection at the wine shop.
  • 3% of respondents wanted a more limited selection at the wine shop.
  • Even those that reported a greater difficulty choosing a wine or those that were pressed for time, still reported being satisfied with the selection at their wine shop.
  • Bordeaux wines were reported to be purchased less frequently and were not as well understood in terms of their taste.
  • Respondents did report an interest in learning more about Bordeaux wines.
  • Respondents also reported an interest in learning more about Argentinian wines.
  • Those wanting to learn more about Argentinian wines expressed interest in a greater selection of Argentinian wines in the shop.
  • Red Zinfandel was reported as not purchased very much.
  • Cabernet Sauvignon-based wines were reported to be purchased in high frequency, though respondents also noted they were not interested in learning more about these types of wines.
  • Pinot Grigio wines were reported to be purchased in high frequency, though respondents also noted they were not interested in learning more about these types of wines.
  • Respondents reported wanting less of a selection of Pinot Grigio wines in the wine shop, even though Pinot Grigio was found to be the largest single category of wine sales at the shops in focus.
  • In a nutshell, no choice overload was determined to exist based upon questionnaire responses.

The Observations

  • Participants self-reported higher wine purchases than actually observed in real-time.
  • Participants self-reported much longer time to shop than they actually did in real-time.
  • Note: questionnaire participants were likely different people than those randomly selected at the store, so there could be differences between the two populations based on that fact alone.
  • 30% of customers were observed to ask for help in the shop (which mostly matched the self-reported questionnaire group at 25%).

The Interviews

  • 119 interviews were performed.
  • Nearly all interviewees reported satisfaction with their wine choice (only 3 people reported less than a 7 out of 10 satisfaction level).
  • The store with the largest selection of wines had the highest overall store satisfaction rate and was tied for the best wine selection satisfaction rate.
  • There was a negative relationship between number of choices and the time it took to select a wine (i.e. the greater the number of wines to choose from, the less time it took to choose a wine).
  • There was no correlation between reported wine knowledge level and satisfaction rate.
  • In a nutshell, no choice overload was determined to exist based upon interview responses.


Overall, the results of this study indicate that choice overload does not exist in the wine retail setting.

Despite these results, I am not too keen to accept them with much confidence. While the concept of the study was certainly fascinating and has the potential to be useful in the wine retail/marketing setting, there are several problems with this study that negate my ability to say with confidence that choice overload does not exist in the general wine retail setting.

First: The sample size was too small. The researchers said it themselves that they needed 843 responses in order to achieve a 95% confidence interval for the questionnaire portion of the study, but received only 806 instead.

Photo by Flickr user John Cooke (

Photo by Flickr user John Cooke (

In order to accept or reject any of the three hypotheses at the 95% confidence interval (the accepted interval for basically all respectable scientific research), you need to have a certain number of study subjects. Based on the numbers crunched by the researchers in this study, that number was 843.

Since they did not achieve this number, the overall power of the study is too low and thereby statistically they are neither able to accept or reject the hypothesis regardless of what results were seen. They say there is no evidence of choice overload based on the questionnaires, however, since their sample size was too low and they did not have enough power, there is no way they can even say that. All they can say is that there is no choice overload for the 806 individuals surveyed, but that’s it.

Second: The participants in the study were not a representative sample of the total American wine consumer population.

In fact, 57% of respondents were categorized as “high frequency wine consumers”. According to a recent Wine Market Council report, only 33% of wine consumers in the US are considered to be high frequency wine consumers, indicating that the sample population for this study was skewed and not representative of the whole.

This means that any of the results of this study cannot be generalized to the overall population, since the sample population does not reflect the general population on a wine purchase frequency level. Even if the sample size were large enough to carry statistically weight, all we could say based on these results is that there is no apparent choice overload noted in high frequency wine consumers in the greater NYC area.

Third: The selection method.

For an exploratory study, the participant selection method was probably sufficient, however, there is likely to be great bias and thus difficulty in extrapolating results to a larger consumer base.

In general, participants were selected based on email lists from nine different wines stores. It’s highly possible that many of those people on that list are already bias toward that one particular store, making discerning the difference between possible choice overload and simple loyalty to that particular store much more difficult.

Selection of individuals from many different types of stores (i.e. supermarkets, wine shops, liquor stores, etc.) or sending people into stores they have never been to or heard of before might help reduce the conflicts of interest in this case.

Overall, I thought the premise of this study was very interesting and would be a valuable addition to the information on choice overload research in general. Taking the researchers at their word, it would seem that choice overload does not exist in the wine retail setting whereas it does exist with other goods.

However, due to the poor power and statistical strength of this study, combined with the poor representation of the true wine consumer population in the United States, I think the jury is still very much out and more research needs to be done on this subject.

What do you all think of this study? Please feel free add to the discussion by leaving your comments below!


Zucker, D. 2015. Drowning in the Wine Lake: Does Choice Overload Exist in Wine Retail? American Association of Wine Economics: AAWE Working Paper #175 (PDF)

12 comments for “How Much is Too Much?: A Study on Choice Overload in Wine Retail

  1. May 28, 2015 at 8:18 am

    As a practicing statistician and market researcher I thought the study results were quite conclusive, The criticisms of the study design are typical from an academic. Marketers do not need to operate with 95% confidence, they make decisions on a day to day basis with far less information. In fact the sample was a good one with a bias to frequent wine drinkers, this was not a concern. These are the target consumers that marketers of wine want not infrequent drinkers. The study confirmed what every market researcher knows – in categories where differences between brands are minor then overload can be a factor, but in categories where every brand has the potential for significant difference then overload is rarely a factor. One quick and easy test would be to run split samples on the data and see if the same results were found. I am sure they would be suggesting some degree of reliability across the whole study. Comments about stores visited as a bias are also strange?. Market studies should map actual behaviour not some randomized design, so buyers commenting on stores they usually frequent is just good design. Despite the misplaced criticisms I found the study results very interesting and indeed logical. There are too many pinot grigio on offer!

    • Becca
      May 28, 2015 at 10:50 am

      Thank you so much for your insightful comments!

      Yes, my background is purely academics and the 95% confidence interval bit was certainly drilled into our brains on an almost daily basis during all the stats classes we took! It’s great to hear from someone with a marketing background!

      Great suggestions on possible additional research/stats options for this study as well.

      Thank you again for reading and commenting!

  2. Z
    May 28, 2015 at 11:15 am

    Flawed study to get anything concrete here – but a good start. Most important was the limited population solicited. Presuming most questioned were regular “large store customers”, you’ve already disregarded the majority of consumers who purchase at the grocery store. All the big wine players know this…

    Second, the details regarding varietals and source countries were too broad to gather anything significant. It is typical for such studies to try to do too much…as was the case here.

    Still interesting to review and build on…


    • Becca
      May 28, 2015 at 7:47 pm

      Great comments! I was also thinking about the grocery store as well—though as Marc mentioned in a later comment not all location (NY for example in this study) allow for wine to be sold in grocery stores, so they would definitely need to reproduce the study in a different state.

      But anyway, I agree—the varietals/source country bit was a little odd and perhaps they took on too much.

      Thanks again for your comments!

  3. May 28, 2015 at 2:41 pm

    Fun and interesting study, but I agree with the points made above. Two other major concerns:

    1) there’s no mechanism reported for ensuring that the observed sample was either a subset of the surveyed wine consumers, or the same type of wine consumer. If so, they can’t make any conclusion about the difference between reported attitudes and observed behavior.

    2) with only three stores selected to represent three different levels of SKU selection – how did the stores differ on other variables (shelf arrangement and information, staff help, merchandising, etc.)? To draw conclusions, you’d want to control for those other variables.

    I don’t understand the researcher’s comment that “843 completed surveys were required for a 95% confidence interval”. You can calculate a 95% confidence interval for almost any sample size, it just gets larger as the sample gets smaller. This needs more context.

    • Becca
      May 28, 2015 at 7:51 pm

      Thanks for your great comments, Christian! You brought up very good concerns and certainly ones that would be important to address in a follow up study.

      In regards to the 95% confidence interval comment, this is what the researchers said exactly:

      “The survey will be randomly sent to 4000 names from the retailers email database. Achieving a representative sample of the population with a 95% confidence level and a confidence interval of 3 will require receiving 843 responses.”.

      My apologies if I was not clear enough in the post about that bit!

      Great comments overall! Thank you for adding to the discussion!

  4. May 28, 2015 at 4:11 pm

    The study is too random just based on geographical location of those interviewed. Wine purchasing on the east coast is much more constrained than on the west coast for example. NY does on allow wine sale in grocery stores. California allows for all wine and spirits to be sold in grocery stores. This fact alone contributes to different buying behaviors and a much larger set of brand choices in any one location in California as compared to New York. I have seen so many consumers staring blankly at “walls of wine” like stunned sheep. If that’s not overload I don’t know what is.

    • Becca
      May 28, 2015 at 7:53 pm

      Excellent points, Marc! I definitely agree with you. I have a feeling we’re definitely not getting the entire picture with just this one study, in part because of the reasons you just described. If it’s economically feasible for the researchers, I would definitely like to see this repeated in different states where sales laws are different from where this study took place to see if and what behaviors change.

      Thanks for your great comments!

  5. May 28, 2015 at 10:08 pm

    Becca this discourse just confirms an interesting but major disconnect between practicing marketers and the academic community. I think any marketing manager would recognize this information as directionally very usable – and as I said a confirmation of what we already believed based on a lot of in-store research.

    All the comments about samples and methodology are really just distractions for in-market practitioners like my clients who are all major brand names. And they are also always the kind of response you get from academia where they feel it is their first duty to challenge research design. Of course everyone would like the “perfect” study, but who can afford that? Its why academic papers based on 100 students always say what further research would be useful and it generally applies to the need for larger samples and more representative design. These kind of comments are basically redundant. What should be focused on is the findings, assuming the research has some credibility.

    Marketers take feedback from any source and anything that has at least some indicators of reliability and validity and design rigour will be accepted as useful in marketing planning. You academics need to get “real” as they say and realize most business decisions are not made on 95% confidence levels or standard errors. And if anyone thinks I don’t know what I am talking about, I am an Econometrician by training and market researcher based on over forty years in the industry. In my view a very useful piece of work

    • May 29, 2015 at 1:14 pm

      I’m not an academic, I’m a market researcher with decades of experience in both research and brand management. My concern about methodology here, specifically the differences in the three stores selected in the observational/interview phase, goes to the heart of the conclusion that choice overload is not a problem. We can’t tell from the article if other store differences or similarities are a factor. Also, all three stores have high # of SKUs (vs. industry averages), so they could all be similar in their choice overload effect. Ergo null hypothesis supported.

      Picky? It doesn’t mean their conclusions are necessarily wrong, and it’s good to have someone revisit this issue. But I’m always cautious if a study contradicts a body of previous research.

      That said, I can certainly relate to your comments on sticking to 95% confidence intervals. I like to joke that even 90% CI is 40% more confidence than the average marketing decision! >:^)

  6. Bob Henry (wine marketer)
    June 8, 2015 at 3:32 am

    Echoing mister8888’s comment, operating in an environment of incomplete information — and using sound judgement to take effective action — is the hallmark of a good business executive.

    (Similarly a good military leader. Consider the phrase “the fog of war.”)

    As for Christian’s comment:

    “I like to joke that even 90% CI is 40% more confidence than the average marketing decision! >:^”

    i infer you are alluding to decision-making by flipping of a coin?

    (Not the oft-quoted 50:50 odds:

    Even worse for spinning a coin.)

    Worth reading: “The Paradox of Choice: Why Less is More” by Barry Schwartz

    Executive summary:

    Worth reading: “Why We Buy: The Science of Shopping” by Paco Underhill

    Executive summary:

    A salient point: In-store sampling converts non-buying shoppers into buyers.

    Quoting from Underhill (page 173, softcover edition):

    “Nobody needs to taste-test Budweiser, but if you’re going to buy that expensive new lambic ale or that Armenian beer, you’ll want to try a little first.” Same goes for fine wine.

    Wineries and their distributors and brokers in the field need to get out to wine stores and food stores, set up sampling tables, and overcome the risk-aversion consumers have in spending money on untried goods.

    And the labels have to convey basic information, according to Underhill (page 142, softcover edition):

    “Why is it that winemakers have begun thinking of their labels as art projects? From Kroger [grocery store] to Trader Joe’s [specialty food store], we’ve documented a kazillion people struggling to read labels. It’s even worse at your local liquor store, where the lighting tends to be dimmer than in the big chains and the shelves can be downright gloomy. I’m not suggesting that a label can’t be pretty or have a kangaroo on it, just that a bottle has to be picked up and glanced at before it gets bought. This is particularly important for small and up-and-coming vintners. Type of wine, country of origin, year [vintage], vineyard and a marketing plug [back label text] — this is all stuff customers are looking for. Proven snotty French brands can do what they want, but all those superb newcomers to the global wine market from Chile, Argentina, South Africa, Australia and New Zealand need to pay attention.”

  7. Bob Henry (wine marketer)
    June 8, 2015 at 3:39 am

    Excerpts from
    (May 12, 2010, 2012):

    “The Market for Fine Wine in the United States”

    [Fine Wine 2010 Conference in Ribera del Duero (Spain)]


    By Graham Holter
    Associate Director – Publishing
    Wine Intelligence market research firm (United Kingdom)

    . . .

    According to the data presented by [David] Francke [managing director of California’s Folio Fine Wine Partners], US wine drinking is compressed into a small segment of the population.

    SIXTEEN PERCENT OF CORE WINE DRINKERS consume wine once a week or more frequently, which ACCOUNTS FOR AROUND 96 PERCENT OF CONSUMPTION. Thirty-five million adults drink virtually all of the wine sold in America, Francke said.

    [Bob’s aside: Corresponds with the “80-20 Rule of Marketing” — 80% of your sales revenue comes from 20% of your customer base. For those more interested in this observed phenomenon, Google these keywords: “Pareto principle” and “Joseph Juran.”]

    . . .

    Wine Intelligence has studied the US wine market in detail and categorised the wine drinking population — which it measures at 47 million — into profile groups. Two of these segments – “Millennial Treaters” and “Experienced Explorers” — were introduced to conference delegates by Erica Donoho, Wine Intelligence’s country manager for the USA.

    “Millennial Treaters,” she said, represent just 6 percent of wine drinkers, but they account for 13 percent of market value.

    “They’re a young group, under 30, and they’re exciting market players to look at,” she said. “Wine was introduced to them at a young age and it’s something they’re embracing wholeheartedly. When we ask them lots of questions, one theme that keeps coming up is there’s a pressure — especially among the men in this group — to know more about wine. They’re receptive to information; they want to be marketed to with some instruction.

    “They’re really interested in sharing knowledge with friends and family, and it’s an amazing way to target this group. They want to share their experience and their knowledge.

    “The social etiquette of wine choosing is becoming increasingly important.”

    Typically, such consumers will use the varietal as a major buying cue, but two thirds of them are also influenced by country or region of origin.

    [Bob’s aside: The article goes on to discuss “Experienced Explorers,” which as a demographic group accounts for 17 percent of the wine drinking population and 33 percent of the market value.]

Comments are closed.