How can an AI Tool Help Watsons to Show Up on Top of Google Shopping Ads?

Challenge​​

Due to limitations from Google Smart Shopping Campaign (SSC), it is the main tool driving conversion in paid media investment (approximately 60-70% of paid media revenue). It maximizes the conversion value effectively through the system’s intelligence calculation, however, the automatic calculation has many limitations. For example, only a few high converting products receive more budget, most of the merchandise share much less budget and cannot get the right ad impressions. SSC automation makes it impossible to optimize manually.

Solution

Unlike Google SSC, which can only handle one budget for the automated calculation and ad delivery, DataFeed+ automatically divides merchandise into default four groups based on past ad performance results.

The optimization can give each group an independent budget so that merchandises that have been buried by the Google SSC and have never been exposed or clicked in the past have the opportunity to appear in front of the consumers. At the same time, the DataFeed+ system will also update each product group in real-time based on the results, allowing the optimizators to have a holistic view of the result and optimize the overall budget allocation at any time. In addition to the four groups preset by DataFeed+, the client can also customize the merchandise grouping rule according to their needs, such as specific performance goal, brand, or product line, so that optimizators can provide flexible strategies for optimization in addition to budget adjustments.

Result

We used DataFeed+ to analyze client data in May 2021 and found that many anti-epidemic related products and main summer products were buried in the system due to fierce bidding competition and Google's algorithm. However, most of the budget was only given to less than 10% of the products in GMC. For delivery, nearly 70% of the products have not appeared in front of consumers or have not been clicked (the actual quantity exceeds tens of thousands). Therefore, we decided to use the DataFeed+ tool for a two-week pilot project starting from June 10th, and automatically segment all merchandise into four groups; especially for the “Ghost” and the “Zombie” product group. We allocated a small amount of budget from the campaign budget, and finally, successfully bring an additional million traffic to the overall campaign and increase ROAS by 95%.