Background

Blibli.com wants to improve the efficacy of digital advertising in a campaign. Blibli, together with Google and Merkle team choose electronics goods as the product with a high transaction for the experiment.

Solution

Blibli, together with Google and Merkle built a model as an approach to predicting the person who has a high propensity to purchase electronic goods in the campaign period from Blibli’s Google Analytics 360 data. The model was created using 90 days of historical data to predict the next 14 days' purchase. The audience is refreshed every week during the campaign period and targets the customers on a high percentile of 90.

Result

The efficacy of digital advertising by utilizing Big Query Machine Learning is 30x better compared to the regular method from Blibli’s campaigns. This initiative also drives 55% more conversion to purchase. The result also showed 30x lower cost Blibli has to spend for one customer to make a single order. 

To better optimize Blibli’s digital advertising campaigns and increase their returns for each campaign spend, they can utilize the Big Query machine learning to create an audience for retargeting campaigns for other categories (i.e. automotive, woman/men's fashion). These audiences proved to give a better result which help Blibli’s team to achieve good returns on every campaign spent from Google Ads.