Data silo problem refers to related datasets located in different databases, systems, or files. In the case of online advertising, performance metrics are fragmented in different campaigns and ad sets, making it difficult to compare customer segments.
In this research, we present SiloSolver, an algorithm that:
- (a) retrieves performance metrics for different customer segments across all campaigns,
- (b) aggregates the values for each customer segment in mean values, and
- (c) runs a statistical comparison (Student’s t-test) on the performance differences between the segments.
The algorithm is evaluated using a real Facebook Ads dataset from an e-commerce company consisting of hundreds of campaigns from over five years.
Using SiloSolver, advertisers using Facebook Ads are better able to understand their market segments across multiple seemingly disparate campaigns.
Salminen, J., Salenius, T., and Jansen. B. J. (2021) SiloSolver: Algorithm for Aggregating Siloed Customer Segments in Facebook Ads Campaigns. The International Conference on Intelligent Data Science Technologies and Applications (IDSTA2021), 15-17 November. Tartu, Estonia. p. 1-5.