• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content

Loading…

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

Like this presentation? Why not share!

How to sell wine online [REPORT]

on

  • 1,016 views

An old market research (2009) about selling wine online. Today it should completely reviewed considering the new role of social media. Anyway it is interesting, especially because of the rigorous ...

An old market research (2009) about selling wine online. Today it should completely reviewed considering the new role of social media. Anyway it is interesting, especially because of the rigorous methodology.

Statistics

Views

Total Views
1,016
Views on SlideShare
1,016
Embed Views
0

Actions

Likes
1
Downloads
8
Comments
0

0 Embeds 0

No embeds

Accessibility

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    How to sell wine online [REPORT] How to sell wine online [REPORT] Presentation Transcript

    • How to sell wine onlineHow to sell wine online Made in Italy S B fi iSusanna Bonafini AnnalauraCantella Cecilia MarchiCecilia Marchi Dario Pagnoni Michele SchincagliaMichele Schincaglia Francesco Travagli 8427 – Market Research
    • AgendaAgenda 1. PHENOMENON AND ISSUES 2. RESEARCH AND BUSINESS GOALS 3. TARGET POPULATION3. TARGET POPULATION 4. SAMPLING: METHOD 5. QUESTIONNAIRE SETTING 6 DATA AUDIT AND OUTLIER DETECTION6. DATA AUDIT AND OUTLIER DETECTION 7. DATA ANALYSIS A. UNIVARIATE B. BIVARIATE C. MULTIVARIATE 1. FACTOR ANALYSIS1. FACTOR ANALYSIS 2. CLUSTER ANALYSIS 3. CONJOINT ANALYSIS 4 DISCRIMINANT ANALYSIS4. DISCRIMINANT ANALYSIS 8. RESULTS 9. MANAGERIAL IMPLICATIONS
    • 1 Phenomenon1. Phenomenon (more) Quality wine is a cultural product that, together with design and fashion, gives h i h “ d i I l ” b d/idemphasis to the “made in Italy” brand/idea . In recent decades, wine has been the center of the enogastronomic tourism boom that madethe enogastronomic tourism boom that made producers grow and offer a complex experiential product: not only wine, but also T l i 2008 th It li t i i d t i fl local culture and values are sold. Truly, in 2008 the Italian enogastronomic scenario made tourism flows grow, with a rise from 18% to 20%. In fact, the visitors interviewed in the tourist place appreciated the Italian, p pp enogastronomic offer so much that gave it an average evaluation of 8.2 out of 10.* *Source: http://www.vinitaly.com/about_news.asp?id=3478
    • Phenomenon (continued) This business doesn’t seem to suffer the crisis: in Italy it involves between 4 and 6,5 million people fond of wine, with a total turnover of about 2 5 billiontotal turnover of about 2,5 billion euro. Italy is the country where Wine & Foody y tourism is most widespread , with 140 “Strade del vino e dei Sapori”. As a result to this growth the sector of wine, which is traditionally static, has opened its doorsy , p to new channels, as online selling. An example is the website Vini24, owned by Sole 24 Ore, and other web portals (e.g. Wine to Wine) based on the collaboration between different quality producersquality producers. Source: http://www.b2b24.ilsole24ore.com/articoli/0,1254,24_ART_96632,00.html?lw=24%3BCHL
    • A glance at Internetg In 2009 Italy placed 20th in a EU l ith I t tEU scale, with an Internet penetration rate of 50.1% compared to the Europeanp p average of 60.7%.1 I l h h I l hIn general though, Italy has been showing record growth in online sales 2in online sales. The main Italian users are young people and well- educated ones, moreover 58 3% of Italians between 1158,3% of Italians between 11 and 74 year old steadily access the web. 2 1 Source: http://www.eiaa.net/news 2Source: own elaboration with ISTAT data
    • What about E-commerce? At the beginning of 2009, GFK Eurisko communicated during a national convention that 5 millions of Italians purchased on the Internet and that they were on average satisfied at 90% of it.1 What’s interesting is that only half of them (52%) hadg y ( ) decided in advance to go online for their shopping, the other half just happened to do it.1j pp This means that starting a business online today has great potentials and maybe it is not so risky as we couldpotentials and, maybe, it is not so risky as we could presume; people trust the Web. Nowadays there’s the possibility to bring online alsoNowadays there s the possibility to bring online also products that we would never think about, for example winewine. 1Source: www.ecommerceforum.it
    • The issues Wine -in particular quality wine- is t diti ll ld th h di t t ttraditionally sold through a direct contact between customer and reseller, who usually gi es ad ises abo t the prod cts andgives advises about the products and transmits its experience. Internet is not so deeply related to the values that this kind of product transmits such asthat this kind of product transmits, such as tradition, culture and natural product. Hence, the main question we want to answer is the following: “I i ibl f d f hi h li i“Is it possible for producers of high quality wine to convey effectively their valuable products through a channel like the I t t? A d f th ti f h tInternet? And, from the perspective of consumers, what are the perceived resentments?”
    • Solution to the problem In order to answer the previous question, we want to analyze some of th t d l t d t li h i f i p the concepts and concerns related to online purchasing of wine. More specifically we will deal with preferences, desires and opinions of the potential customers.potential customers. For our purpose, we need to: Develop a market research on the potential target population Understand who our respondents are and how they behave in terms of: • Wine preferences and purchasing habits • Internet usage and online purchasing This way we will try to understand if the online channel could bey y operable for an experiential product as wine.
    • 2 Research Goals2. Research Goals 1 • Identify the potential market for online wine selling• Identify the potential market for online wine selling 2 • Point out the preferences of the potential consumers• Point out the preferences of the potential consumers 2 p pp p P i t t th t t f th t ti lP i t t th t t f th t ti l 3 • Point out the resentments of the potential consumers• Point out the resentments of the potential consumers 4 • Segment the demand• Segment the demand 5 • Identify the most important potential segments• Identify the most important potential segments
    • Business Goal We will take the point of view of wine producers that want to challenge Internet as a distribution channel, therefore we will use the information gathered from the research to develop the right strategy in order to hit h diff t t ill id tifeach different segments we will identify. This task will comprise setting: 1 • PRODUCTS and SERVICES to be offered • PRODUCTS and SERVICES to be offered 2 • PRICING• PRICING 2 3 • PROMOTION• PROMOTION 3 4 • OPTIMAL WEBSITE structure and d i • OPTIMAL WEBSITE structure and d i4 designdesign
    • 3 Target population3. Target population • O l i t fi d b i l ti f th it li k t Italians • Our goal is to find a business solution for the italian market Between 18 and 64 • This is the most interesting age range of potential buyers of wine online Between 18 and 64 wine online Internet users • In order to buy wine online, obviously people should use Internet • Even if there could be people buying wine without drinking it, Wine drinkers E e t e e d be pe p e b y g w e w t t d g t, we consider only wine drinkers
    • 4. Sampling- Stratified Sample l l p g p 1st STEP In order to stratify the M F TOT Italian population 2008 y sample, we divided the Italian population into 18 - 24 2.197.942 2.099.564 4.297.506 25 - 34 4.077.971 3.999.081 8.077.052 35 - 44 4 895 311 4 840 448 9 735 759 strata according to the gender and the age range. 35 44 4.895.311 4.840.448 9.735.759 45 - 64 7.559.637 7.853.523 15.413.160 TOT 18.730.861 18.792.616 37.523.477 Source: ISTAT Italian Internet users 2008 2ND STEP W f d th b f M F TOT 18 - 24 1.460.328 1.412.807 2.873.135 2 34 2 3 06 2 039 31 4 396 99We found the number of Internet users for the selected strata 25 - 34 2.357.067 2.039.531 4.396.599 35 - 44 2.486.818 1.965.222 4.452.040 45 - 64 2 682 192 1 558 190 4 240 382selected strata. 45 - 64 2.682.192 1.558.190 4.240.382 TOT 8.986.405 6.975.751 15.962.156 Source: ISTAT
    • Sampling - Stratified Sample 3RD STEP Italian drinking wine 2008 p g p 3RD STEP Here, we made a basic assumption: the distribution of wine drinkers within the M F TOT 18‐24 48% 31% 39% g Italian population remains stable also within the Italian sub-population that uses Internet 25‐34 65% 41% 53% 35‐44 72% 44% 58% Internet. It means assuming that, among the people who use Internet, we will find as many i d i k th h l It li 45‐64 79% 48% 63% TOT 67% 41% 54% wine drinkers as among the whole Italian population. Using Internet & drinking wine Starting from the absolute numbers of step 2 and applying the percentages of wine M F TOT 18‐24 694.412 435.398 1.129.810 drinkers among the Italian population using Internet, we found the number of Internet users that are also wine drinkers. 25‐34 1.524.533 832.972 2.357.505 35‐44 1.793.045 865.661 2.658.706 45‐64 2.123.595 744.822 2.868.417 TOT 6.135.585 2.878.852 9.014.437
    • Optimal sample sizeOptimal sample size Considering our “budget” and time available, we thought that the best size of our sample was 200.p Italians using Internet & drinking wine (percentage) M F TOTSo we proceeded Italians using Internet & drinking wine (percentage) 18‐24 8% 5% 13% 25 34 17% 9% 26% p by finding the percentage of 25‐34 17% 9% 26% 35‐44 20% 10% 29% percentage of Internet users that are also wine 45‐64 24% 8% 32% TOT 68% 32% 100% are also wine drinkers. TOT 68% 32% 100%
    • Optimal sample size (2) In this way we found the exact number of males and females, for Optimal sample size (2) each age group, that were necessary for our research and that perfectly reflected reality.
    • 5 Questionnaire5. Questionnaire We structured our questionnaire dividing it into 4 sections: 1. Wine consumption 2. The usage of the online channel 3. The online purchase of wine 4. Personal and behavioral data We sent more than 300 questionnaires through the web, trying toq g y g reach different target groups (considering age, gender and social status). Within the collected questionnaires we extracted randomly the exact number of males and females of all age groups that wasexact number of males and females, of all age groups, that was required by our sample size of 200.
    • Questionnaire PrizeQuestionnaire - Prize Internet-based questionnaires usually have a particular disadvantage: the lowest response rate among any other method of ll ti d t I d t id thi bl d t t t h thcollecting data. In order to avoid this problem and try to catch the attention of the interviewees, we decided to do a prize-winning lottery:lottery: • Our respondents could leave their email dd t th d f th ti iaddress at the end of the questionnaire, in order to participate; • On Dec. 10 we randomly drew 3 emailOn Dec. 10 we randomly drew 3 email addresses from the list mentioned before • The winners were contacted through h i il dd d h dtheir email addresses and the rewards were delivered in few days • The other participants were thankedThe other participants were thanked through an email
    • Respondents’ Profile - OriginRespondents Profile Origin O li t h i t th t b f l d f lOur sampling technique set the accurate number of males and females per age range. In order to better understand who are our respondents we analyzed their personal data:analyzed their personal data: Origin Even though our aim was to consider people from all the areas of Italy, the majority of our respondents turned out to come from the North. We canj y p address this to research limitations. Th f d h d ‘E ’ I li ifThe four respondents that answered ‘Estero’ are anyway Italians – even if originally coming from abroad – and so they are part of our target.
    • Respondents’ Profile - EducationRespondents Profile Education Educational LevelEducational Level The results regarding the educational level are quite significant: most of them d l d t h hi h h ldeclared to have an high school diploma and a slightly lower percentage declared to have apercentage declared to have a degree. Hence, the majority of the respondents have a quite high d ti l l l G SC OO D P O A DEGREE educational level. HIGH SCHOOL DIPLOMA MIDDLE SCHOOL DIPLOMA
    • Respondents’ Profile - Profession Profession p The profile of the respondents that concerns the profession shows a high partition of the sample between different fieldsbetween different fields. We can say that the sample is quite equal subdivided amongequal subdivided among self-employed, employee, students and “other”. Just few of them declared to be housewives or pensioners. SELF-EMPLOYED HOUSEWIFE PO P S O REMPOYEE PENSIONER STUDENT OTHER
    • Respondents’ Profile- Hobbiesp 120 140 80 100 40 60 0 20 Considering the fact that the respondents had more than one possible choice, the distribution of hobbies is quite homogeneous among commonchoice, the distribution of hobbies is quite homogeneous among common activities like Sport, Literature and Music. We will see next that there are no statistically significant differences in terms of hobbies and behaviors among our segments, which confirms the fact that wine is appreciated by everyone.
    • Respondents’ Profile - Preferred Holidaysp y When asked about holidays, it is evident that the majority of theevident that the majority of the respondents chose an adventurous trip on the road,p which has almost nothing in common with wine. But quite interesting, as we will see in the following slides is thesee in the following slides, is the fact that the portion of people that chose Wine & food tourism Total relax On the road Wine&food tourism has a higher degree of wine consumption per month. Wine&food tourism
    • 6 Data audit6. Data audit Before starting the analysis of the data we obtained, it is fundamental t h k if th it bl f th l i Thi fi t f llto check if they are suitable for the analysis. This means, first of all, that we have to consider all the missing cases and blank values, and then decide what to do with themdecide what to do with them. Since our research is considering Internet users, we chose to submit ti i th h li l tf i d t tour questionnaire through an online platform, in order to create an initial filter (respondents are, for sure, Internet users). This platform doesn’t show in the final output all the questionnaires that were abandoned by the respondents, hence we have no missing cases. On the other hand, it was possible for the interviewees to leave some ofp the multiple choices, that’s why we do have some blank cells.
    • Outlier Detection Before proceeding with the data analysis we decided toBefore proceeding with the data analysis we decided to investigate if all the numerical answers to the open questions of our questionnaires presented extreme cases or outliersour questionnaires presented extreme cases or outliers. – In case of extreme cases we substituted them with the average value of the variableaverage value of the variable. – In case of “acceptable” outliers we decided not to change h d ifi h lidi dthem and not sacrifice the content validity and accuracy. The process is shown by the example given by the outlier detection of Question 2: “How many bottles of wine do you consumedetection of Question 2: How many bottles of wine do you consume at home within a month?”
    • Outlier Detection - Results Before Respondents to questionnaires n. 23 and 17 declared23 and 17 declared to consume respectively 300espectively 300 and 100 bottle of wine. They represent Extreme Cases- b bl i tprobably input errors- and we decided todecided to substitute them with the average value, i.e. 9.
    • Outlier Detection - Results After We ran again the outlier analysisoutlier analysis and we found out that the respondent to the questionnaire ld bn. 46 could be considered an “acceptable”acceptable outlier and though was leftg unchanged in the data set.
    • 7 Data Analysis7. Data Analysis A. Univariate: F i & Ch t-Frequencies & Charts -Descriptive Statistics B. Bivariate: - Analysis of connections - Analysis of correlation - ANOVA C. Multivariate
    • Univariate Descriptive Statistics We performed the univariate descriptive statistics in p p p order to better understand who are our potential customers and how is their purchasing behavior: Wi f s d h si g h bits1 p g Wine preferences and purchasing habits1 Internet usage and online purchasing2 Preferences in buying wine online3 y g3 Resentments in buying wine online4
    • Wine preferences and purchasing habits 1 p p g 1 WHAT? After a brief description of the respondents, we want to d d h b h i f h i h iunderstand the behavior of them with respect to wine. When asked which wine type they liked, our Red Wine respondents had more than one possible White Wine Spumante answer, and this is the final distribution of Rosè their preferences: red wine is the most frequent type.
    • Wine preferences and purchasing habits WHERE? Winer Manufacturer Supermarket Others Wine Bar Online 0% 5% 10% 15% 20% 25% 30% 35% 40% The most frequent places where the respondents usually buy wine are directly the prod cer and the s permarkets this co ld mean that they are eitherthe producer and the supermarkets, this could mean that they are either interested in quality or they don’t care and choose simple supermarkets. The online channel obtained less than 5% of answers, which tells us that we do have, some room to take in the sector.
    • Wine preferences and purchasing habitsp p g WHEN? 160 180 200 100 120 140 160 60 80 100 0 20 40 D ti ti I di id l ti R t t ith th R t t l A ift When asked for which occasions they purchase wine, the respondents Domestic consumption Individual consumption Restaurant with others Restaurant alone As a gift had more than one choice, and the most frequent answers were “Domestic consumption” and “Restaurants with others”. The results show that they usually enjoy wine when they are in good companyshow that they usually enjoy wine when they are in good company, such as with friends and family members.
    • Wine preferences and purchasing habits HOW MUCH ( th)? Observing the Mean Value we can see that our respondents consume approximately 8 HOW MUCH (per month)? g p pp y bottles of wine per month with friends and family, while they buy approximately 2 bottles of wine per month for gifts. We can notice that the Standard Deviation is pretty high, showing an high variability in the consumption habits of the sample. The skewness is positive inp both wine consumption and wine gifted, this because of a long right taillong right tail. From the kurtosis we can d t d th t th di t ib tiunderstand that the distribution of both the variables is more peaked than a normal one.
    • Wine preferences and purchasing habits WHY?WHY? Descriptive Statistics N Range Minimum Maximum Mean Std. Deviation Variance Wine type 198 9 1 10 8,07 2,460 6,052, , , IGT/DOC/DOCG certification 197 9 1 10 7,39 2,586 6,687 Manufacturer reputation 198 9 1 10 6,89 2,218 4,922 Geographical origin 198 9 1 10 6 58 2 583 6 671Geographical origin 198 9 1 10 6,58 2,583 6,671 Alcohol gradation 197 9 1 10 4,69 2,433 5,921 Label attractiveness 196 9 1 10 4,23 2,245 5,039 V lid N (li t i ) The chart shows which are the major characteristics that guide consumers while Valid N (listwise) 194 j g choosing their wines and which are strictly related to wine. The higher mean value outlines that the most important factor is the type of wine and then the quality certification (IGT, DOC, DOCG), the reputation of the producer and the geographicalcertification (IGT, DOC, DOCG), the reputation of the producer and the geographical origin of the wine. Less important characteristics are the alcohol gradation and the label attractivenessLess important characteristics are the alcohol gradation and the label attractiveness. Standard deviations are similar for all the characteristics.
    • Wine preferences and purchasing habits WHY2?WHY2? DescriptiveStatistics N Range Minimum Maximum Mean Std. Deviation Variance Only known manufacturers 194 9 1 10 5,64 2,842 8,075 Only wines already tasted 195 9 1 10 4,94 2,810 7,899 Only wines advised by confidants 196 9 1 10 4,59 2,556 6,531 Onlywinesadvisedbyguides/magazines/oth ers 196 9 1 10 3,98 2,374 5,636 Only wines on sales 194 9 1 10 3,09 2,386 5,691 Only wines from my region 196 9 1 10 2 71 2 129 4 533Only wines from my region 196 9 1 10 2,71 2,129 4,533 Valid N (listwise) 189 The data indicate the reasons that guide consumers through the choice of the wine. The majority of them prefers to choose wine bacause of the reputation of the producer and they also prefer wines that have already tasted or for which they were advised. This reflectsy p y y the need of wine consumers to ensure the quality of wine they are buying. Anyway, in order to make comparison with statistical significance, we should wait for theAnyway, in order to make comparison with statistical significance, we should wait for the bivariate analysis (the same worth also for the previous slide and for other similar slides following in this section).
    • Internet usage and online purchasing P i d i k2 -Perceived risk-2 The graph shows the di t ib ti f ti th tdistribution of rating that our respondents gave for the perceived risk ofp online purchases: the average value (4,72 out of 10) i t hi h10) is not very high. This information gives us the confidence to say thaty Internet is a real potential channel for our business, as we had mentioned inas we had mentioned in the introduction section of this work.
    • Preferences in buying wine online W b it f t3 -Website features- hi h h b i h i i h ld ff d 3 Which are the website’s characteristics that could affect our respondents’ purchases of wine online? According to the mean values, the two most important ones are the ease of the purchasing and payment process and theimportant ones are the ease of the purchasing and payment process and the ease of seeking a specific wine within the website list. Lower importance is given to the spotlight on rebate and layout of the website; hi l b i f d l hi h l f d d d i ithis last website feature presented also an higher value of standard deviation than the first ones. DescriptiveStatistics N Range Minimum Maximum Mean Std. Deviation Variance Purchase and payment processes ease 200 9 1 10 8,58 1,648 2,717 Ease to seek a specific wine 200 9 1 10 8,36 1,672 2,794 Feedbacks on desired wine 200 9 1 10 7,30 2,215 4,905 Spotlight on rebate 200 9 1 10 6,43 2,305 5,311 Layout & graphics 200 9 1 10 6,11 2,397 5,747, , , Valid N (listwise) 200
    • Preferences in buying wine online Additional services-Additional services- The following data refer to the overall characteristics that most influenceThe following data refer to the overall characteristics that most influence consumers while purchasing wine online. By looking at mean values, the most important aspects are the availability of special wines and shipping convenience from the website, immediately followed by the return policies. Our marketing strategy will probably stress these services, improving and advertising themadvertising them. Standard deviations are pretty similar and relatively high. DescriptiveStatistics N Range Minimum Maximum Mean Std. Deviation Variance S i l i ' il bilitSpecialwines' availability 200 9 1 10 7,47 2,112 4,461 Shipping convenience 200 9 1 10 7,40 2,385 5,688 Return policies 200 9 1 10 6,72 2,966 8,796 Professional advises 200 9 1 10 5,68 2,862 8,190 Community and blog to exchange feedbacks 200 9 1 10 5,50 2,597 6,744 L b l t i tiLabel customization 200 9 1 10 4,24 2,852 8,133 Valid N (listwise) 200
    • Resentments in buying wine online -Service characteristics-4 -Service characteristics- A really important aspect of our analysis is to understand the possible 4 resentments of the customers about wine online purchases. The service that less induces consumers to buy online is the absence of human contact and the second one is the scare not to receive the wine purchased Both these negativesecond one is the scare not to receive the wine purchased. Both these negative aspects are related to the concept of security we noticed also for the wine purchase behavior, as mentioned previously talking about wine purchasing attitude. The possibility that other people could not appreciate customers’ choice of wine doesn’t seem to influence negatively consumers’ behaviordoesn t seem to influence negatively consumers behavior. Descriptive Statistics Std. N Range Minimum Maximum Mean Deviation Variance Human contact 200 9 1 10 6,94 2,951 8,706 Scare to not receive the wine 200 9 1 10 5,47 2,841 8,070 Scare to buy a wine that dislike 200 9 1 10 5,16 2,874 8,262 Delivery time too long 200 9 1 10 5,02 2,607 6,798 Scare to make a choice that others could mock 200 9 1 10 2 60 2 182 4 763mock 200 9 1 10 2,60 2,182 4,763 Valid N (listwise) 200
    • Bivariate Descriptive Statisticsp In order to understand if there is any association between the selected variables, we performed the bivariate analysis through different tools: Analysis of Connection: Contingency tables1 Analysis of Connection: Contingency tables1 Analysis of Correlation: Scatter plots and tables2 Analysis of Variance: Means by classes3 y y3
    • Analysis of Connection1 y1 This analysis investigates the dependency between two categoricaly g p y g variables through cross tabulations which show counts and percentages of the variables’ combinations. The variable used are, for example: Gender, Range of age, Demographics, Yes/No questions. We analyzed many combinations of variables that we thought would be important, but most of them weren’t significant. For Example: - Age ranges and Online purchases (YES/NO) - Profession and Online purchases (YES/NO) - Age ranges/Gender/Profession and Online wine purchases (YES/NO) A possible cause of the absence of relationship between variables could be the low number of respondents per each category of the variables considered.
    • Age ranges & O li h Fascia_età * Online purchaseCrosstabulation Online purchase Si No Total Online purchases (YES/NO) Fascia_età 18-24 Count 14 11 25 % within Fascia_età 56,0% 44,0% 100,0% % within Online 16 1% 9 7% 12 5% purchase 16,1% 9,7% 12,5% % of Total 7,0% 5,5% 12,5% 25-34 Count 22 30 52 % within Fascia_età 42 3% 57 7% 100 0% SymmetricMeasures Value Approx. Sig. 42,3% 57,7% 100,0% % within Online purchase 25,3% 26,5% 26,0% % of Total 11,0% 15,0% 26,0% pp g NominalbyNominal Phi ,151 ,207 Cramer's V ,151 ,207 NofValid Cases 200 Cramer’s V index signal some 35-44 Count 29 30 59 % within Fascia_età 49,2% 50,8% 100,0% % within Online 33 3% 26 5% 29 5% 200 statistical dependence between the “Fascia d’età” and “Online purchase”, but the Approx Sig is too high: the purchase 33,3% 26,5% 29,5% % of Total 14,5% 15,0% 29,5% 45-64 Count 22 42 64 % within Fascia età but the Approx. Sig. is too high: the overall dependent relation between the variables is not significant. % within Fascia_età 34,4% 65,6% 100,0% % within Online purchase 25,3% 37,2% 32,0% % of Total 11 0% 21 0% 32 0% Therefore, we will present only the % of Total 11,0% 21,0% 32,0% Total Count 87 113 200 % withinFascia_età 43,5% 56,5% 100,0% % within Online 100 0 significant and satisfactory analysis of connection… % within Online purchase 100,0% 100,0 % 100,0% % of Total 43,5% 56,5% 100,0%
    • Age ranges & Preferred payment modalities (online) The following statistics assess the dependency between the age variable (range) and the preferred payment method. Chi-Square value is higher than 0 and the C V i d i hi h th 0 1 Chi-Square Tests Asymp Sig Symmetric Measures Value Approx Sig Cramer V index is higher than 0,1. Value df Asymp. Sig. (2-sided) Pearson Chi-Square 2,453E1 12 ,017 Likelihood Ratio 27,382 12 ,007 Value Approx. Sig. NominalbyNominal Phi ,350 ,017 Cramer's V ,202 ,017 NofValid Cases 200 Linear-by-Linear Association 10,369 1 ,001 N of Valid Cases 200 a. 6 cells (30,0%) have expected count less than 5. The minimum expected count is ,88. The cross tabulation (in the output document) shows the frequencies of respondents’The cross tabulation (in the output document) shows the frequencies of respondents preferred payment modality with respect to their age. Two main trends can be detected: -Young/middle age people prefer Prepaid Card and Paypal(quite new payment tools)g g p p p p yp (q p y ) - Middle/older age people prefer Credit Card and Bonifico(more “traditional” ones)
    • Analysis of Correlation2 Analysis of Correlation2 This analysis aims to determine if there is a significant linear relationshipy g between two quantitative variables such as Monthly wine consumption, Minutes spent to purchase wine, Age, Number of online purchases, Willingness to pay, etc. As we did with the previous analysis, we combined many variables which couldp y , y be significant, if correlated, but most of them weren’t statistically significant. For example: - Age and Willingness to pay for private consumption/gift/restaurant - Willingness to pay for a gift and Monthly wine gifted - Monthly wine gifted and Minutes spent to purchase wine for gift - Number of online purchases and Acceptable delivery price Th f ill p t l th l i f ti l t fTherefore, we will present only the analysis of connection relevant for our marketing goals…
    • Monthly wine consumption & M thl i ift dMonthly wine gifted The correlation between the variables is significant (the p- value is 0), although the R2 isvalue is 0), although the R is low. The number of bottles of i d d h dwine consumed and purchased for gifts within a month are positively correlated.positively correlated. (See also the following slide) This information could be used to address specific promotions to the “loyalp p y customers” that buy lot of bottles online.
    • Age & Monthly wine i d ifconsumption and gift Correlations Age Monthly wine consumption Monthly wine gifted Age Pearson Correlation 1,000 ,250** ,191** Sig. (2-tailed) ,000 ,007 N 200 000 200 200 The correlation between wine purchasing &N 200,000 200 200 Monthly wine consumption Pearson Correlation ,250** 1,000 ,531** Sig. (2-tailed) ,000 ,000 purchasing & consumption and age is slightlyg ( ) , , N 200 200,000 200 Monthly wine gifted Pearson Correlation ,191** ,531** 1,000 g g y positive and significant. Sig. (2-tailed) ,007 ,000 N 200 200 200,000 **. Correlation is significant at the 0.01 level (2-tailed). This information tells us that as the age of our respondents increases also wine consumption does. Older people could therefore be more likely to buy wine more often.
    • Analysis of Variance3 Following, two analysis of variance. We examined many other variables but we didn’t l y3 find any interesting values in term of significance and Eta Squared. Monthly wine consumption & Preferred holiday type As we could imagine, the monthly wine ti i hi h f l th t y p y yp consumption is higher for people that declare to prefer an enogastronomic holiday.y Even if the Eta Squared is low -and so the strength of the effects of the preferredEven if the Eta Squared is low and so the strength of the effects of the preferred holiday type on the monthly wine consumption is weak- the ANOVA table shows that the p-value is significant. It is therefore possible to conclude that the difference between means of consumption of wine with respect to the 3 types of holiday is significant.
    • Perceived risk for online purchasesp Respondents who experienced online purchasing rated Report Perceived risk for online purchases p p p g lower (almost 1 point) the perceived risk to buy online. Since the number of respondent who has never purchased online is really high the issue of the p Online purchase Mean N Std. Deviation Si 3,91 87 2,009 purchased online is really high, the issue of the perceived risk should be taken into consideration when developing the business. No 5,34 113 2,408 Total 4,72 200 2,348 ANOVA Table Sum of Squares df Mean Square F Sig.q q g Perceived risk for online purchases * Online purchase BetweenGroups (Combined) 100,269 1 100,269 19,923 ,000 Within Groups 996,486 198 5,033 Total 1096,755 199 The average risk perceived among people who purchased online and who didn’t is significant. Measures ofAssociation p g Eta squared is low, so the strength of the effects of having purchased online and the perceived risk of Eta Eta Squared Perceived risk for online purchases * Online purchase ,302 ,091 having purchased online and the perceived risk of online payment methods is weak.
    • Multivariate i i i iDescriptive Statistics The multivariate statistic techniques allow us to go more in depth in the analysis of our respondents inmore in depth in the analysis of our respondents in order to reach the research goals Factor Analysis1 Cluster Analysis2 Cluster Analysis C j i A l i 2 3 Conjoint Analysis3 Discriminant Analysis4
    • Factor Analysis1 In order to start the behavioral segmentation of our respondents and then classify them into different segments, we chose to run a factor analysis of some variables: they were too correlated with each other and would have increased the complexity of our work. Therefore, we selected four questions that we were particularly interested in,Therefore, we selected four questions that we were particularly interested in, and managed to reduce their number without losing significance. The questions were: n. 21, n.22, n.23, n. 8: ─ Layout & graphics E h ifi i ─ Scared to buy a branded wine that I don't like H i i d h l b l h h ld─ Ease to choose a specific wine ─ Spotlight on rebate ─ Feedbacks on desired wine P h d t ─ Hesitation towards the label that others could mock ─ Scare not to receive the wine E t l l d li ti─ Purchase and payment ease ─ Special wines' availability ─ Shipping convenience Return policies ─ Extremely long delivery time ─ Human contact ─ Origin Gradation (alc%/vol)─ Return policies ─ Return policies ─ Community and blog to exchange feedbacks ─ Gradation (alc%/vol) ─ Producer ─ Label ─ Typefeedbacks ─ Professional advises ─ Label customization ─ Type ─ Certification
    • First try: error The first time we ran the analysis, we weren’t satisfied with its output. In fact, the factors we obtained were affected by three major problems: • “1/3 Th ” Th t t l b f i bl i d i th l i (22)• 1/3 Theory : The total number of variables comprised in the analysis (22) divided by 3 didn’t give the number of factors (6) • The percentage of explained variability was too low (59.14%) Percentage of explained variability p g p y ( ) • The Scree plot was not satisfactory Percentage of explained variability
    • What did we change?What did we change? At this point we decided to go through all the variables of theAt this point, we decided to go through all the variables of the selected questions, to understand which ones were the most interesting for our purpose.g p p Through this process we eliminated some of them, because they weren’t related to our research as the others. These variables are: • n 8 1 (Origin)• n. 8_1 (Origin) • n. 8_2 (Gradation: alc%/vol) • n. 8_5 (Type)_ ( yp ) • n. 23_2 (Hesitation towards the label that others could mock) • n. 23_5 (Human contact)
    • Second try: ok! (more)y Without the variables listed here theWithout the variables listed here, the output of the analysis was satisfactory under all the aspects (see next slide): • “1/3 Theory”: The total number of variables comprised in the analysis (17) di id d b 3 i t l thdivided by 3 gave approximately the number of factors (6) • The percentage of explained variabilityThe percentage of explained variability was high enough (66,61%) • The Scree plot was not quite satisfactory • The Eigen value was fine (1,01) • The quote of overall variability explained by each output was high • The correlation structure was well d fi ddefined
    • Second try: ok! (continued) The following table shows precisely the information explained in the previous slide y p
    • Optimal factors These factors comprise all the concepts we are most interested in, which are: the elements people consider when purchasing wine, the external influences (being them human or related to the brand) and the additional services that could affect the purchasing process. Rotated Component Matrixa Component Website features Additional services Delivery Aesthetics External advises Quality Layout & graphics 740Layout & graphics ,740 Ease to choose a specific wine ,690 Spotlight on rebate ,670 Purchase and payment ease 615 322 316Purchase and payment ease ,615 ,322 ,316 Feedbacks on desired wine ,538 ,426 Shipping convenience ,852 Special wines' availability ,704 Return policies ,589 ,361 Scare not to receive the wine ,845 Extremely long delivery time ,759 Label customization ,698 Scared to buy a branded wine th t I d 't lik ,407 ,698 that I don't like Label attractiveness -,311 ,598 Community and blog to exchange feedbacks ,850 Professional advises 324 652 328Professional advises ,324 ,652 ,328 IGT/DOC/DOCG certification ,768 Manufacturer reputation ,692
    • Description of final factors Factor 1: Website f Website components that affect the consumers when purchasing a product and that could improve or worsen the p features process. Factor 2: Additi l Services that could add value to the core product, and that Additional services Services that could add value to the core product, and that normally make the difference when building loyalty. Factor 3: Delivery Usual concerns of consumers when choosing Internet to purchase products. Factor 4: Aesthetics Subjective reactions when purchasing an experience good like wine, they are mainly related to brand image and social status.y y g Factor 5: External Opinions and suggestions that could influence consumers’ External advises p gg choices. Factor 6: Quality Elements that usually assure the actual quality of wine.
    • Cluster analysis2 Once obtained the factors, we were able to run a cluster analysis aimed to classify our consumers into relatively homogeneous groups, based on the factorized set of variables considered. Being our sample relativley big, we ran the cluster analysis with the k-means method.y But, before doing it, we used the hierarchical methodg on a sample of 61 cases to identify the optimal number of clusters and initial seeds to run the k-means.
    • Hierarchical with sampleHierarchical with sample After observing the dendrogram we conclude that good numbers ofAfter observing the dendrogram, we conclude that good numbers of clusters could be either 3 or 6.
    • Final choice: 3 clusters ( )Final choice: 3 clusters (more) Considering both the analysis of the dendrogram and the nature of our market research, we decided that 3 could be a good number of clusters (3 segments of customers is enough for an online business). Hence, we saved the initial seeds found through the hierarchical cluster l i th lanalysis on the sample. Then, we used the initial seeds to run a k-means analysis on all the cases, considering as cluster number 3considering as cluster number 3.
    • Final choice: 3 clusters As we expected after deciding a number of clusters relatively lower than Final choice: 3 clusters (continued) As we expected after deciding a number of clusters relatively lower than the number of factors, one of these turned to be not very significant for the analysis.y In particular , Website Features seems to be not very relevant for thepa ticula , Website Features see s to be ot ve y eleva t fo the creation of our clusters.
    • Final choice: 3 clustersFinal choice: 3 clusters (continued) The analysis of the ANOVA suggests to increase the number of clusters. Anyway though, the dendrogram suggests a solution of 3 or 6 clusters. Considering our marketing objectives and the nature of our business, we believe that a 3 clusters are still the best solution. So, finally, we found 3 relevant l h ill b d ib d iclusters, that will be described in the following slides.
    • Cluster analysisCluster analysis - Final cluster centers - The final cluster centers table shows a significantshows a significant characterization
    • General description of clustersGeneral description of clusters • Insensible to additional services, external i fl s d d li o ditio s Th Quality seekers influences and delivery conditions. They just want the best quality. Q y (cluster 1) • Slightly sensible to additional services, theyLow involved Slightly sensible to additional services, they are not emotionally involved in the wine purchasing process Low involved (cluster 2) • Favorable to the services connected to buy li d ibl i d hWeb confidents online and sensible to community and other external advices. They also care about quality. Web confidents (cluster 3) q y
    • Detailed description of clustersp In the next slides we will analyze our segments in a more detailed way, iny g y order to try and see how they differ from each other and thus how we can hit them with our marketing strategy. This is aimed to avoid the problem of the current online wine resellers, that don’t know who their customers are. Analyzing the main demographical and behavioral information (e.g. hobby, preferred holiday type), we observed that there are no statistically significant differences among the three clusters we identified. This is probably due to the fact that, in Italy, wine is a traditional element, it is part of our culture. Both young and older people enjoy it, no matter what their life habits are. If we were dealing with other markets, like the United States we probably would have found less people that drink wineUnited States, we probably would have found less people that drink wine as often as Italians do, and in fewer occasions. Anyway there something that differentiate the clusters we will discussAnyway, there something that differentiate the clusters; we will discuss it in the following slides.
    • Cluster 1: Quality seekers M F TOT 18-24 25-34 35-44 45-64 TOT Sample size 29%M F TOT 68% 32% 100% 18 24 25 34 35 44 45 64 TOT 7% 30% 24% 39% 100% As in the other clusters, the percentage of women is about half of the percentage of men. This is also because of the characteristics of the target population (italiansmen. This is also because of the characteristics of the target population (italians drinking wine and using Internet) for the selected strata. The majority of the people is between 45 64 years old and only very few are young F h t i i th i ttit d Q lit k i l i t t d i The majority of the people is between 45-64 years old and only very few are young (18-24). So, as we could presume, Quality seekers tend to be older. For what is concerning their attitudes, Quality seekers are mainly interested in purchasing the best quality of wine, and therefore pay significantly more attention to: -Wine certifications (IGT/DOC/DOCG)( ) - Wine producer’s reputation On the other hand, they consider less important additional services offered by the website.
    • Cluster 2: Low involved M F TOT 18-24 25-34 35-44 45-64 TOT Sample size M F TOT 72% 28% 100% 18 24 25 34 35 44 45 64 TOT 21% 15% 25% 39% 100% 20% Thi i th l t ith th hi h ti f l b t dThis is the cluster with the higher proportion of people between 18 and 24 year old. It is also the smallest cluster. The peculiarity of the Low involved is that they are not concerned about particular aspects related to wine and website They didn't gaveabout particular aspects related to wine and website. They didn t gave significantly high rating of importance to any of the question we asked, showing low interest and involvement in the topic.g p As it’s very clear especially from the Radar graph, they are not concerned at all about quality elements, but consider a lot potential ddi i l i ff d b h b iadditional services offered by the website.
    • Cluster 3: Web Confidents M F TOT 18-24 25-34 35-44 45-64 TOT Sample size M F TOT 67% 33% 100% 18 24 25 34 35 44 45 64 TOT 13% 27% 32% 28% 100% 51% The age distribution of the clusters' members show that the majority of them are middle-aged: 59% of them are between 25 and 44 years old. l h l h h b f d hBeing also the cluster with the bigger size, Web confidents seem to be the right people to target. This cluster is the most interesting one among all the others, because it gave high ratings of importance to wine and website features. With respect to the other clusters, they are really sensitive to:p , y y - Special wines availability - External advises (feedbacks on blogs and professional advices) Moreover, the average of their preferences is quite high for almost all the variables. H t p di t hi h l t p b l t ?How to predict which cluster a person belongs to? We will see it with the discriminant analysis, after the conjoint analysis…
    • Conjoint analysis3 j y We decided to use a conjoint analysis in order to better understand th f f l t d diff t ibl t tthe preferences of our sample toward different possible structures of the wine online purchase. We selected a set of attributes and levels represented as follows: Discount We selected a set of attributes and levels represented as follows: Payment: Delivery: Discount •10% per ordini superiori a 20 Payment: • Carta di credito/prepagata Delivery: •1 giorno lavorativo superiori a 20 bottiglie • 10% i t credito/prepagata • Contrassegno t l lavorativo • 2-3 giorni lavorativi • 10% su acquisto successivo ogni 3 ordini postale •PayPal lavorativi • 4-5 giorni l i i ordini • Sconti sui vini del mese proposti lavorativi mese proposti
    • Conjoint analysis - ResultsConjoint analysis Results With th tt ib t d th ibl l l f h fWith three attributes and three possible levels for each of them, spss figured out 9 possible product configurations, hi h h b l t d b lwhich have been evaluated by our sample.
    • Output InterpretationOutput Interpretation • The most relevant attribute is the payment method: 49 17• The most relevant attribute is the payment method: 49,17 • In particular, the prepaid/credit card system (“carta di credito/ prepagata”) positively influences the global judgement of the selling proposition, while Cash-on-delivery (“contrassegno postale”) and PayPal impact less because of their minor importance values. Analyzing the utilities of our respondents we can figure out the best scenario: 1 Rebates on Credit/prepayed card 1 Day- delivery Rebates on monthly suggested winesy gg
    • Conjoint Gender To evaluate how the perception of the attribute varies according to the personal Classification variable data of the interviewee, we chose to run again the analysis considering demographical variables: Gender, Age and Educational Level.
    • Output InterpretationOutput Interpretation -Gender- • Comparing the results of the importance values we find no big differences between males and females, however it’s possible to note that the payment method attribute remains the most important for both genders, even if with slightly different weights. • Comparing the utility results we note that females perceived Credit card payment more important than menperceived Credit card payment more important than men did. • On the other hand, for males the presence of discounts on the monthly suggested wines has -slightly- more importance.
    • Conjoint Age (18-34)j g ( )
    • Conjoint Age (35-64)j g ( )
    • Output InterpretationOutput Interpretation -Age- • Comparing results, it’s clear that our segmentsp g g consider “payment” the most important component. We want to highlights that this attribute is particularlyg g p y important for people between 45-64. • Moreover, the same age class perceive respect theMoreover, the same age class perceive respect the other age classes the “delivery” attribute (12,392) as the less important.the less important. Ed i l L lEducational Level We run the conjoint also splitting for the educational levels, but we found that both the utilities and the importance values were similarfound that both the utilities and the importance values were similar among the different educational levels.
    • Linear discriminant analysis4 Linear discriminant analysis4 “Often you have measured different groups of respondents on i i bl Di i i l i i f lmany metric variables. Disriminant analysis is a useful way to answer the questions... Are the group different? ...On what i bl h diff ? C I di hi hvariable are they most different? ...Can I predict which group a person belongs to using these variables?” Jamie Baker-Prewitt,, vice President, decision sciences,, Burke Inc.
    • Linear discriminant analysisLinear discriminant analysis Do the different market segments (clusters) differ in their structure of preferences about the characteristics? • Independent variable market segments (clusters)• Independent variable: market segments (clusters) • Dependent variables: – Geographical origin – Alcohol gradationAlcohol gradation – Producer’s reputation L b l tt ti– Label attractiveness – Wine type (red, white, sparkling, Rosé) – IGT/DOC/DOCG certification
    • Linear discriminant analysisLinear discriminant analysis • The valid cases considered in the analysis are 194 in total• The valid cases considered in the analysis are 194 in total • Just 1 case lacked the evaluation in at least one of the discriminant variables (missing data)
    • Linear discriminant analysisLinear discriminant analysis • Only the variable “Geographical origin” is not significant Th t i ifi t i bl i th “IGT/DOC/DOCG• The most significant variable is the “IGT/DOC/DOCG certification”
    • Linear discriminant analysisLinear discriminant analysis The Eigen values show that the first function explains the 92,4% of the variance, while the second one explains the residual 7,6% Obviously, the coefficient of canonical correlation Eta is higher for the first function. Looking at Wilks’ Lambda, it is notable the significance of the two discriminant functions.
    • Linear discriminant analysisLinear discriminant analysis 1° Function: certification, producer’s reputation and wine type 2° Function: label attractiveness, alchool gradation and geographical i iorigin Rotated pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions . Variables ordered by size of correlation within function. *.Largest absolute correlation between each variable and anyg y discriminant function
    • Li di i i t l iLinear discriminant analysis With the rotated structure matrix and the functions at group centroids we can create a map, in order to better understand the findings of the discriminant analysis. 1°function X axis wine quality indicators °f i Y i l f ifi d f2°function Y axis relevance of specific product features
    • Discriminant analysis findingsDiscriminant analysis findings Do the different market segments differ in their structure of preferences about the wine characteristics? C di hi h l b l ?Can we predict which cluster a person belongs to? • We analized the discriminating power of the preference towards different characteristics of wine • With a brief questionnaire -that should be submitted to the website k h f f h h d husers- asking the preferences for the wine characteristics used in the analysis, we can presume the cluster each user belongs to • In this way, we can address to each client the right selling proposition and marketing mix according to his cluster membershipand marketing mix, according to his cluster membership
    • Linear discriminant analysisLinear discriminant analysis Wine Characteristics Clusters Label attractiveness Quality seekers IGT/DOC/DOCG certificationcertification Wine type Geographical origin Low involved Manufacturer reputation Alcohol gradation Web Confidents X axis Wine quality indicators Y axis Relevance of specific product features
    • Linear discriminant analysisLinear discriminant analysis Wine Characteristics Clusters Label attractiveness Quality seekers IGT/DOC/DOCG certificationcertification Wine type Geographical origin Low involved Manufacturer reputation Web & wine lovers Alcohol gradation Web & wine lovers X axis Wine quality indicators Y axis Relevance of specific product features
    • 8 RESULTS FINDINGS) P t ti l k t f li i lli 8. RESULTS FINDINGS) Potential market for online wine selling Th t ti l k t f b i i it h t iThe potential market for our business is quite heterogeneous in terms of gender, age, origin or profession: from one side, Internet is accessible basically to everyone; from the other wine drinkersis accessible basically to everyone; from the other, wine drinkers don’t seem to have peculiar characteristics. We detected just positive correlation between age and wineWe detected just positive correlation between age and wine consumption (per month) and a slight relationship between age and preferred payment method online.and p efe ed pay ent ethod online. The online channel is not very common for wine selling (less than 5% of our sample declared to experienced it) but it has goodp p ) g potential because 43,5% of respondents used Internet for purchasing. Moreover the risk perceived for the online purchasing in general is rated 4,72 (out of 10), so relatively low.
    • FINDINGS) Preferences of the potential consumers (more)) p ( ) Preferred wine type: considering that it was a multiple choice answer, the distribution of preferences in quite homogeneous among red, white and ( h h d h li k d) Th i i ’spumante (even though red was the most clicked). The wine type it’s also the major characteristic that guide consumers while choosing their winestheir wines. Anyway, before taking decision, a view of the trend along the time should be consideredbe considered. Most frequent consumption occasions are meals at home or restaurantMost frequent consumption occasions are meals at home or restaurant, always with other people (family or friends), but also for gifts. People use to drink almost 8 bottles per month and purchase for gifts almostp p g 2, in average.
    • FINDINGS) Preferences of the potential consumers (continued) When choosing wine, people give importance to the wine type, but also quality certifications, producer’s reputation and advises of trusted people. ) p ( ) p p p p The last element could be interesting if we consider the potential of web 2.0 instrument. Interaction among users, but also between the (trusted with good reputation) firm, will be very important. W ld l i i i l t k f i t f “Wi b k”!We could also imagine a social network of wine, a sort of Winebook ! Respondents of our questionnaire who prefer food & wine tourism consumeRespondents of our questionnaire who prefer food & wine tourism consume significantly more wine (per month) so the association between these two variables could be exploited for communication and advertising. And what about online wine purchasing preferences? - People declared very important the ease of purchasing process and of seeking a specific wine. Th l hi h t f i t t th il bilit f i l i- They also gave high rates of importance to the availability of special wines, the shipping convenience from the website, return policies.
    • FINDINGS) Resentments of the potential consumers) p Our respondents assess that their major resentments regarding Internet channel are: • the absence of human contact • the scare not to receive wine purchased B h f hi ifi f h i d bBoth of this resentments are not specific for the wine product category but reflect the general distrust attitude toward the online world. So it seams that there aren’t, for wine, significant more resentments than in other, , g industry. This is also a matter of “generational consumption”. We can presume that, with the passing of time, resentments will decrease. Anyway it is very important to build a trusted brand well known to theAnyway it is very important to build a trusted brand, well known to the costumers.
    • FINDINGS) Demand segmentationFINDINGS) Demand segmentation W i d f b h i l i b did ’ b iWe tried to perform a behavioral segmentation but we didn’t obtain a satisfactory result , maybe also because our sample was quite little. Moreover, we realized that there are no particular elements that differ wine drinkers inp Italy. Thus we decided to perform a demographical segmentation but again ended up with poor results. Finally we chose to perform a segmentation using significant factors, and realized that it was the best solution since this method really comprisesrealized that it was the best solution, since this method really comprises what’s important for our business. Through the factor analysis we obtained 6 factors that summarize all the aspects of purchasing wine online (quality, aesthetics, external advises, website features, delivery, additional services) and according to them we created 3 clusters, called: Web confidents, Quality seekers, Low involved.
    • FINDINGS) Potential and most important segmentsFINDINGS) Potential and most important segments We analyzed a sample representative of the Italians that are wine drinkers and Internet users, but it is unrealistic to think that all of them would buy winey online. Therefore, considering our three clusters, we think that efforts should be, g , addressed to hit cluster n. 1 and 3: Quality seekers and Web Confindents. Even if they are different in terms of internal preferences and behaviors, they l th t i ifi t d ibl t i d b h t i tiare also the most significant and sensible to wine and web characteristics. Afterwards we will give suggestions to create the best online platform possible, in order to attract the consumers that are looking for an innovative and easy way to get their quality wine. We will not consider to invest money on Low Involved people.
    • Limitations of our analysisLimitations of our analysis Before moving on with our final consideration on the managerial i li ti f thi k ti h t t t iimplication of this marketing research, we want to repeat once again the limitation of our analysis: • Time constraints • Budget limitationsBudget li itations • Questionnaires have been distributed to relatives and friends, trying also to reach more people from different areas of Italy But intrying also to reach more people from different areas of Italy. But in the end the majority of people turned out to be from the North of Italy anyway.taly a yway. • This is our first marketing research!
    • 9. Managerial Implications9. Managerial Implications The best business idea Starting form the analysis of our data, we want try to outline some of the characteristics of the online selling that would match better with The best business idea the characteristics of the online selling that would match better with the segments identified, according to their habits, preferences and needs. We are conducting this analysis of the managerial/marketingWe are conducting this analysis of the managerial/marketing implications from the point of view of : • PRODUCTS and SERVICES offered • PRICINGPRICING • PROMOTION • OPTIMAL WEBSITE structure and designOPTIMAL WEBSITE structure and design
    • PRODUCT and SERVICES Everyday offer and Special wines In the website, consumers should find the same assortment of a real wine cellar, especially because one of our clusters is really fond of wine (Quality seekers) and we want to make sure that they trust the high standard of itseekers) and we want to make sure that they trust the high standard of it. Anyway a selection of specific wines should be created and proposed togetherAnyway, a selection of specific wines should be created and proposed together with advices of experts. Therefore we suggest to rely only on experienced and well-known producers, to propose a wide offer that comprises both everyday and special occasions’ b lbottles. I dditi i ti l t ( Ch i t l h bi thd fIn addition, in particular moments (e.g. Christmas, seasonal changes, birthday of customers) some peculiar wines should be offered, in order to enhance loyalty or just please the clients.y y j p
    • PRODUCT and SERVICES Ease of finding wines One of the first behavioural output of our analysis is that the majority of the respondents considers as the starting point of the decisional process the type of wine they want to buy (white, red, sparkling, Rosé) and, moreover, one of the most important characteristics of the website is the f ki ifi iease of seeking a specific wine. Considering these two outputs, the website should make a clear di i i f diff idistinction of different wine types: in this way customers will feel guided through the choice and will be helpedthrough the choice and will be helped in order to find the desired product without wasting time searching on theg g website.
    • PRODUCT and SERVICES Feeling of human contact and experience One more evident result of our analysis is that the majority of our respondents prefers to choose the wine basing on the reputation of the producer. F thi d t il d i f ti b t th d h ld b id dFor this reason detailed information about the producers should be provided and, moreover, it should be possible to contact them directly: in this way costumers will feel free to ask them questions regarding the product and willq g g p solve all doubts they have with regard to their purchase. Moreover we suggest to give the opportunity to get real backups from a sommelier, who will share its experience through a dedicated blog. A id b f ld l i i i l k f i fAs said before, we could also imagine a social network of wine, a sort of “Winebook”, a Facebook of wine!
    • PRODUCT and SERVICESPRODUCT and SERVICES Payment process Since the prepaid/credit card system positively influences the global judgment of consumers, and since one of the characteristics of the website that have a strong impact on consumer behavior is purchase and payment ease, we strongly believe that the website has to be organized in such way that a safestrongly believe that the website has to be organized in such way that a safe payment is granted. Moreover, the process doesn’t have to be too complicated: our data show that this attribute is particularly important for people between 45-64 and we don’t t t t b k th h f thi t t th t ld b l kill dwant to set back the purchases of this target group that could be less skilled with the use of technologies. Hence, we propose to set an easy process, that grants security and that allow payments through prepaid and credit cards, in order to satisfy both young and older people.
    • PRODUCT and SERVICESPRODUCT and SERVICES Delivery and return policies From the analysis resulted that our respondents prefer the shortest delivery time possible, that is one working day. The only ones that didn’t worry a lot about this aspect were people between 45-64 year old. Moreover, the clusters we chose to address showed to be quite interested also about return policies – if you buy more than one bottle of the same wine but, while you open the first one, you don’t like it, you can return the unbroachedwhile you open the first one, you don t like it, you can return the unbroached bottles for free. Thus, our suggestion is to engage a reliable shipping company to grant the safest delivery of wine, and to offer return policies to manage situations of damaged b ttl ti fi d tbottles or unsatisfied customers. This will also appease those consumers that are scared about not receiving theirThis will also appease those consumers that are scared about not receiving their purchases, and that don’t fully trust Internet.
    • PRICING Unfortunately, our marketing research doesn’t allow us to gain relevant consideration about the pricing strategy. For sure, from one hand, the online channel allow to cut costs. In this way it would be possible use a strategy of lower prices.p gy p On the other hand, wine use the pricing also a as a signal for the quality: a lower price could be perceived as a lower quality. So, the basic price should be the same of the traditional channel. But it would be ibl k i l ipossible to make special promotions. M i i h i it f th b it t ifi l t it ld bMoreover, assigning each visitors of the website to a specific cluster it would be possible to adopt a customized pricing strategy for him/her.
    • PROMOTIONSPROMOTIONS O l i tli th t th b f i h i thOur analysis outlines that the number of wine purchaser increases among those people who prefer an Wine and Food tourism. So, sponsorships and co-marketing activities with travel agencies or holiday farms could be for sure a success, bringing advantages both to the website and to the other company. Often tourists visit a region and fall in love with the local food and wine. If we promote to them the wine website, they will be very happy to order and continue drinking wine also from they home!drinking wine also from they home! In addition, we suggest to cooperate with some restaurants: if they purchase from, gg p y p the website cellar, the website could be advertised on their wine selection menu (“Carta dei vini”).
    • OPTIMAL WEBSITE DESIGNOPTIMAL WEBSITE DESIGN In conclusion, we would like to point out some of the elements that an optimal websitep p should have to potentially and successfully sell wine. Si l d i t ti b l t• Simple and interactive web-layout • History section dedicated to wine producers information and contacts • Website section dedicated to special wine & offering (promotions discounts and freeWebsite section dedicated to special wine & offering (promotions, discounts and free deliveries ) • Website section called “Talk to the Sommelier” where consumers can share information d b i h l lk di l i li d li fand comments about wines; they can also talk directly to a specialized sommelier for advices and tips • Website section dedicated to all the events linked to wine and enogastronomicg experiences all over Italy • Web tool to make an easy research of different wines based on wine’s name, producer’s name origin wine type and all these attributes combinedname, origin , wine type and all these attributes combined • Web tool, such as shopping carts and personal account, to make an easy purchase and payment processe and always keep track of personal orders, payment and delivery status of the product purchased
    • CONCLUSIONSCONCLUSIONS “It is well to remember that there are five reasons for drinking wine:f f g the arrival of a friend, one’s present or future thirst, the excellence of the wine, or any other reason” L i bLatin proverb “Wine is a bottled poetry” Robert Louis Stevenson