Managing Large Data Flashcards
- The correct management of large portions of data has become an emergent but challenging issue for companies. This issue becomes incredibly complex in omnichannel retailing, particularly when it comes to the problem of Data Quality and Data Integration. Based on the paper of Verhoef et al. (2010), explain the reasons why these two processes remain critical as a challenge for companies trying to operate seamlessly across several channels.
Data quality and data integration are extremely difficult for multichannel retailers, which collect customer data across multiple channels. Ideally, data should be integrated across channels to provide a complete view of customer activity and facilitate one-to-one marketing.
Data comes in with different formats, from different databases, and uses different reporting standards. This, as different entities and providers acquire information through diverse touch-points. Also, the reliability of sources differs, as their providers can have different reliability levels, as they are located inside and outside the firm.
All in all, retailers with online channels have the opportunity to gather data that is potentially richer than traditional customer data, as customers’ online browsing behavior can be followed extensively. These data are often referred to as clickstream data. Customers can be followed during their buying process, from entering the website to finalizing the sale.
Data quality and data integration are extremely difficult for multichannel retailers, which collect customer data across multiple channels. Ideally data should be integrated across channels to provide a complete view of customer activity and facilitate one-to-one marketing. Research has shown that data quality is positively related to performance, although some argue that the marginal benefits of improved data quality will decrease while costs increase. This implies that the profit-maximizing level of data quality will be realized with not-perfect data. Despite their insights, there is limited empirical evidence on the optimal level of data quality and data integration. Moreover, cost of data quality and integration may not rise continuously, but may increase in a stepwise fashion. It is also important that different functions, such as marketing and service, have access to the same data.
- Based on Li and Kannan (2014), what is a common limitation of last-click attribution metrics and seven-day average metrics concerning crediting different channels related to conversion outcomes in a customer’s purchase funnel?
The authors suggest that neither the last-click attribution metrics nor the seven-day average metrics are good measures for understanding the real impact of firm-initiated channels nor customer-initiated channels on conversions. The problem rely on the fact these metrics consider only those visits that result in conversion immediately.
Although they may provide mediocre results in product categories with a very short purchase decision hierarchy (e.g., with one or two touch-points) and with fewer channels, they will invariably be misleading in product/service categories with a more extended purchase decision hierarchy, as in high-involvement categories (e.g., consumer durables, travel services), as well as for firms with multiple channels, both customer and firm initiated.
In the latter case, it is also expected that the last-click model would underestimate the effectiveness of firm-initiated efforts.
In sum, this question also requires using the student’s paper’s information more analytically.
It is related to the common limitations of last-click attribution metrics and seven-day average metrics. The answer is given in the student’s paper: “previously used metrics (last-click and 7-day-average) underperform in explaining the influence of diverse channels on conversion”. This initial answer should be complemented with more information, which is also given in the student’s paper (p.9), to derive that these two metrics fail when, for instance, a conversion funnel has more than one touchpoints or when the consumer’s journey has an offline portion.
These are the two most widely used metrics in the industry for determining the contribution of each channel to purchase conversions:
(1) The last-click attribution metric gives all credit to the visit at which conversion occurred.
(2) The seven-day average attribution metric assigns the conversion credit equally to all the visits made in the previous seven days.
Limitations of last click attribution: it ignores the prior channel touches. For example, a customer might decide to buy the product when he visits their website, so the last click attribution will attribute that conversion to the website. However, this person has visited the store multiple times and saw several display banner ads of this product on Youtube.
→ Last click attributes the conversion solely to the website while the other channels played a role in the conversion as well.
→ Carryover and spillover effects of other channels are ignored
Common limitations of last-click and seven-day average attribution metrics: These metrics consider only those visits that result in conversion immediately. Although they may provide passable results in product categories with a very short purchase decision hierarchy (e.g., with one or two touchpoints) and with fewer channels, they will invariably be misleading in product/service categories with a longer purchase decision hierarchy, as in high-involvement categories (e.g., consumer durables, travel services), as well as for firms with multiple channels, both customer and firm initiated.
