Bazaar.com is a leading online retailer in the US with extensive advertising presence in both display advertising and search engine advertising running paid search ads on Google and Bing. Bazaar’s paid ads are broadly classified into two main categories by keywords: branded and non-branded. Brand keywords contain the ‘Bazaar’ brand name such as ‘Bazaar’, ‘Bazaar shoes’, ‘Bazaar clothes’ and so on, while non-branded are keywords such as ‘shoes’, ‘dress’ that do not contain ‘Bazaar’.
With regards to traffic data by Google and Bing, the marketing analytics team at Bazaar computed a 320% ROI on their sponsored ad spending. This result is skeptical because people who search with the word ‘Bazaar’ already had the intent to visit Bazaar.com, hence the effectiveness of branded keyword ads is implausible. Our goal is to understand the true causal effect of the search ads and their effectiveness.
We conducted Difference in Difference(DiD) to examine the causal inference and recalculated the true ROI.