Product unavailability is a common problem for grocery shoppers. Traditional research has shown that stock-outs have a negative impact for retailers, both directly (on category sales and profit) and indirectly (on customer satisfaction, store loyalty and retail image) (Campo, Gijsbrechts and Nisol 2003; Fitzsimons 2000; Sloot, Verhoef and Franses 2005). Recent evidence suggests that out-of-stock (OOS) problems are not limited to traditional supermarkets, but may constitute a far more daunting problem for e-grocers – who experience more severe forecasting problems and strongly fluctuating demand (Fitzsimons 2000). According to a recent survey, product unavailability also rates second in the top 3 of online shopping irritations (Marketing online 2004).

Despite their importance, stock-outs have not been investigated systematically in an online grocery context. Yet, such research could be insightful: differences between the virtual and traditional environment possibly causing divergent online and offline OOS reactions. Moreover, online OOS research is especially valuable as the virtual store environment offers unique opportunities to alleviate the negative OOS effects. Whereas traditional grocers typically leave the OOS item’s shelf space empty, online grocers have the possibility to pursue OOS policies that are not or difficult to implement in a traditional store. First, while stock-outs in brick-and-mortar stores are visible for all customers, online retailers can use the virtual store environment to make OOS failures less prominent. Specifically, they may choose to notify only buyers of that particular product, an approach we refer to as the ‘non-visible’ policy. Second, an online retailer can, compared to a traditional retailer, more easily compensate for the OOS inconvenience, using a technical process that immediately suggests a replacement item each time an OOS occurs (a ‘replacement’ policy). The main objective of this study was to compare the effectiveness of these two alternative OOS policies with the traditional OOS approach (a ‘visible, no replacement’ policy) on online grocery category purchase and choice decisions.

Data were collected by means of a realistic online shopping experiment, where respondents could make purchases during six fictitious weeks for two product categories (margarine and cereals). The net sample comprised 584 respondents (response rate of 17%). In spite of many differences between online and offline stores, OOS reactions in online grocery stores appear to be very similar to reactions observed in traditional stores. Most consumers prefer to buy another item, rather than dropping the category purchase or visiting another store.

Our results further suggest that the adopted OOS policy significantly affects category purchase incidence as well as choice decisions. First, we find that a non-visible policy (where consumers only become aware of a stock-out when they click on the product to purchase it) creates confusion and intensifies the consumer’s loss experience, thereby reducing the tendency to buy in the category. Especially consumers who tend to repurchase the same item react negatively: they clearly prefer to know the real assortment they can choose from upfront. Second, we find that suggesting a replacement item may dramatically increase the consideration and (hence) choice probability of the suggested item. An important implication is that retailers can re-direct choices in case of stock-outs, by suggesting appropriate substitutes. Yet, these positive replacement effects only prevail under certain conditions. For one, the positive effect disappears when higher-priced items are suggested. In such cases, consumers find the recommendation ‘suspicious’ – ascribing it to retailers’ self-serving motives – and do not adopt the replacement item. Conversely, consumers value the replacement policy especially when the suggested items have a low degree of consideration (i.e. are not already salient) yet high intrinsic utility (i.e. appeal to the consumer). While the replacement policy strongly affects consumers’ choice decisions, it hardly influences the probability that consumers will cancel the store purchase. Future research should verify whether this also holds in more complex choice situations, where consumers are more easily inclined to give up planned purchases in case of OOS disruptions.

Taken together, the results demonstrate that grocery e-tailers should carefully select their OOS policies. On the one hand, they should make sure to avoid policies that raise ‘suspicion’, such as hiding stock-outs or suggesting higher-priced options. On the other hand, online retailers can strongly guide consumers’ choices in an OOS situation by adopting a replacement policy with properly selected (maybe less salient, but intrinsically similar) replacement suggestions. Overall, our results indicate that online grocery retailers have an interest in pursuing open and convenience-oriented OOS policies. In line with customer relationship marketing principles, such a customer-oriented OOS approach will benefit both the retailer and consumer.