Influence of pull marketing actions on marketing action effectiveness of multichannel firms: A meta-analysis

We use the terms “distribution channel” and “channel” interchangeably. Channel, in this study, refers solely to distribution channels and not communication channels such as media outlets. Although communication is certainly possible via distribution channels such as by using display ads within online websites, we take the perspective of distribution.

We reviewed meta-analytic studies published in major marketing journals up until 2021 to confirm that we have sufficient data to conduct a meta-analysis. The number of articles used in previous meta-analytic studies ranges from 9 (e.g., Tellis and Wernerfelt 1987) to 1,030 (e.g., Peterson 1994), and the number of effect sizes varies from 11 (e.g., Cox et al., 1997) to 11,874 (e.g., Homburg et al., 2012). We thus conclude that our research contains an adequate number of studies for a meta-analysis.

Although there are many types of digital advertising, studies mostly investigate one or the other. Due to the large missing values of every single type of digital advertising, we are unable to separately consider them theoretically and empirically. As a result, we club everything under the umbrella of digital advertising.

We recognize that the opposite is less likely to happen such that showrooming customers (if they can be tracked in physical stores, which is difficult in itself) are less likely to purchase online from the same multichannel firm (Forbes, 2016).

These journals include the Journal of the Academy of Marketing Science, Journal of Marketing, Journal of Marketing Research, Journal of Interactive Marketing, Journal of Retailing, Journal of Marketing Channels, Marketing Science, International Journal of Research in Marketing, Management Science, Journal of Service Research, Marketing Letters, Quantitative Marketing and Economics, Journal of Retailing and Consumer Services, Journal of Business Research, European Journal of Marketing, Information Systems Research, MIS Quarterly, Strategic Management Journal, Decision Sciences, Decision Support Systems, Industrial Marketing Management. Many other journals were although not considered as major marketing journals publishing multichannel distribution-related work. Several such journals did not turn up in many of our electronic databases but did turn up on Google Scholar. For example, we considered studies in Journal of International Marketing, International Journal of Retail and Distribution Management, to name a few. We included region-specific journals as well such as Asia–Pacific Journal of Marketing and Logistics as they showed up on Google Scholar. We note that these studies form a preliminary sample, and many studies were eliminated based on the inclusion criteria that we describe next.

Marketing actions encompass all types of 4 P elements or marketing-mix deployed by multichannel firms.

This happens when the authors employ different modeling techniques. For example, in Breugelmans and Campo (2016), we collected 72 MAEs measuring how a multichannel firm’s price promotion affects its business performance. The researchers investigated this IV–DV relationship based on 12 samples of different customer segments, product types (e.g., milk vs. cereal), and modeling techniques (i.e., homogenous vs. heterogeneous models). For each sample, we extracted six MAEs measuring the cross-channel (online and offline) price promotion effects (i.e., contemporaneous, lagged, and frequency). In a similar vein, Gu, and Kannan (2021) employed three versions of DID to investigate the heterogeneities of the treatment effect of mobile app adoption on customers’ multichannel spending. The authors set up 10 groups among six customer segments for comparison (i.e., Segment 1 vs. 3, 4, 5, and 6 respectively; 2 vs. 4, 5, 6 respectively, 3 vs. 5, 6 respectively, and 4 vs. 6). Eventually, we collected 30 MAEs (10 groups $$\times $$ 3 modeling techniques) from this study.

We should also control for the total number of channels in the distribution portfolio. However, the number of channels correlates substantially with channel variety (.71), which precludes us from using the number of channels as a control in the analysis.