Respondents were categorized into benefit segments based on their responses to the two shopping motivations scales (goal-orientation and experiential orientation) described in the appendix. A rigorous multi-step cluster analysis method was used, and is described in detail in the appendix. The resulting segmentation of e-shoppers is presented in Table 1.
Table 1 about here
Cluster 1 is the largest (45.5%) of the three. This cluster was labelled “goal-oriented e-shoppers” on the basis of a high mean in the goal-oriented measure and the respective low value of the mean in the experiential scale. The second cluster, which was the smallest, (20.5% of the e-shoppers) was labelled “e-shoppers with mixed orientation” as the two means in the two scales were average. Finally, the third cluster, which represented one-third of the e-shoppers, was labelled “experiential” e- on the basis of the value of the two means compared to the whole sample mean. Findings conform to the work by Wolfinbarger and Gilly, which reports that goal-oriented shoppers constitute the largest segment.
Table 2 about here
Table 2 highlights the demographics and other differences across the three shopper segments (using chi square test for gender and one-way ANOVA for the other variables). Most of the previous segmentation studies (for example, Westbrook & Black; Rohm & Swaminathan) failed to identify any demographic differences on benefit shopper segments. Interestingly, Bellenger and Korgaonkar suggested that experiential shoppers are more likely to be female. Indeed, in our sample, proportionally there did seem to be more women than men in that segment and fewer women in the goal-oriented segment. However, the relationship could only be accepted at a level higher than that the conventional significance level of 0.05. Bellenger, and Korgaonkar, also suggested that experiential shoppers differ from the other groups on adopted lifestyles. Lifestyle may be a more effective variable than demographics to identify and segment the three categories of shoppers.
There are no notable differences regarding the main (dependent and independent) variables of interest. The only exception was perceived within customisation of the site, where goal- oriented shoppers recorded higher scores than the other segments. This may have to do with the choice of sites by this segment and is in line with the argument leading to hypothesis H1c.
Regression Analysis: Testing of the hypotheses
Hypothesis H1, H3a and H4 related to the influence of e-tailer differentiation strategies (as seen by customers) on satisfaction that were tested through ordinary least square regression analysis. To identify the moderating effect of benefit-segment membership, analyses were conducted for each of the three shopper segments, and the whole sample as well (Table 3). The last column of Table 3 reports the significance levels of the interaction between the two segments associated with the hypotheses (i.e., goal-oriented and experiential shoppers) and each of the differentiation strategy covariates (segment x covariate) included in an ANCOVA design (the two shopper segments is the fixed variable, the independent variables are the covariates and satisfaction the dependent variable). Demographics were not included in any of the regression equations, as they did not have any significant effect on the dependent variables. Their inclusion would have added only noise and reduced the power of the analysis.
Results provided empirical support for H1a, H1b and H1d. As expected, convenience and customer care had a significant effect on satisfaction of the goal-oriented shoppers. Value for money (H1d) also had an effect on the satisfaction of goal-oriented shoppers. However, as can be seen in Table 3, value for money also contributes to the satisfaction of experiential shoppers. However, the significance levels reported in the last column of Table 3 suggest that the effect of value for money to satisfaction is higher in the goal-oriented shopper segment than the experiential shopper segment.
The most interesting finding was related to customisation. Contrary to what was hypothesised in H1c, customisation seems to lead to higher levels of satisfaction in the experiential shoppers segment than in the goal-oriented one. This relationship needs to be re-examined in other contexts, in relation to the information-seeking propensity of experiential shoppers. Given the fact that goal-oriented shoppers had higher scores in customisation (see Table 2) than the other types of shoppers, it may be interpreted that it is not a determining variable for the satisfaction of that segment.
Regarding experiential shoppers (H4), only hypothesis H4b was confirmed. Indeed, product assortment strategies are more appealing in terms of satisfaction to the experiential shoppers segment. However, differentiation based on website character (H4a) and product quality (H4c) were not found to have any effect on the satisfaction of that or any other segment. Similarly, interactivity (H3a) did not contribute to the satisfaction of any of the segments. The significance levels in the last column (labelled Interact) of Table 3 suggest that only the regression coefficients for customisation, customer care, convenience, product assortment and value for money were statistically different across the two segments.
Table 3 about here
To test hypotheses H2, H3b and H5, hierarchical regression analysis was used for each shopper category. Satisfaction was entered in the first block, followed by the independent variables and satisfaction in the second block. As mentioned earlier, satisfaction is closely related to loyalty, and it was necessary to decompose the variance of loyalty explained by satisfaction and the other variables. Similar to Table 3, the last column of Table 4 reports the significance level of the interaction terms (segment × covariate) in the ANCOVA design described earlier. Consistently, with the previous sections only the interaction effects of the two key shopper segments were examined.
Findings in Table 4 suggest that independent variables could explain an important part of loyalty in both the goal-oriented and experiential shopper segments. Specifically, the change of R2 (ΔR2) attributed to the variables of interest was 12% for the goal-oriented shoppers segment and 31% for the experiential shoppers segment. Collectively, the independent (differentiation strategy) variables explained a higher portion of loyalty in the experiential shoppers’ segment than the goal-oriented shoppers’ segment. Surprisingly, satisfaction explained 47.1% of loyalty variance in the goal-oriented cluster and only 27.8% in the experiential cluster. The statistical insignificance of satisfaction in the experiential shopper segment suggests that satisfaction does not necessarily lead to loyalty in that segment. It seems likely that the motive of this segment of shoppers compels them to try new experiences in different sites, regardless of the level of satisfaction that they receive from current sites. This implies that managers can influence the loyalty of experiential shoppers directly (by offering certain features) rather than by trying to increase satisfaction. Satisfaction monitoring and generation strategies would be more effective for the goal-oriented shoppers segment. This difference is confirmed by the observed significance level in the last column (entitled “interact”) of Table 4.
Only hypothesis H2d, which postulated that value for money differentiation would increase loyalty of the goal-oriented shoppers’ segment, was confirmed (Table 4). As expected, (H2e) product quality had a direct effect on the loyalty of the goal-oriented shoppers but not on that of the experiential shoppers. This is interesting, because product quality did not have any impact on the satisfaction of any of the segments (Table 3). However, the interaction effect between shopper segment loyalty and product quality is not statistically significant. None of the other variables had a direct effect on loyalty. As was established in Table 3, customer care and convenience had an indirect effect on loyalty (mediated by satisfaction). The mediating role of satisfaction to the two strategies was also statistically confirmed through a Sobel test.
|
