Success story.
From manual calculation to 66,313 elasticity models: prices decided with data
A consumer goods company used to calculate price elasticity by hand: a slow, poorly scalable process without a solid analytical basis for deciding its pricing.
We take on any challenge
At Zenta there are no obstacles we cannot overcome.
Challenge
Automate and migrate the calculation of price elasticity to Google Cloud Platform, replacing the manual process with a metric capable of estimating the elasticity of a product over a given period.
Solution
Using a traditional seven-stage methodology, we built the infrastructure on GCP and automated data ingestion from Cloud Storage to Dataflow and BigQuery. In Python, we developed the cleaning and processing pipeline and promotion detection (lift and shift using the client's business logic). After exploratory analysis (ANOVA, correlations, and bivariate analysis) to detect patterns and anomalies by subchain and product, we applied a frequency-weighted regression model, calculated elasticities, and validated performance using RMSE, MAE, AIC, and BIC.
Results from our management
66,313 statistically significant, validated, and well-fitting models.
Regular and promotional price elasticities calculated at the room-product level using multiple regressions.
Elasticities at the sub-chain-product level obtained with a calculation weighted by the sales volume of each room.
Fully automated calculation in GCP, without relying on the manual process.
Solid and scalable analytical foundation for designing pricing strategies and planning effective promotions by SKU.
Stage 1 of the project successfully completed thanks to the joint work of the client's teams and Zenta Group.