Supply chain · Production

Cainz cuts sales data preprocessing time to 50 minutes across 209 stores with Vertex AI

The problem

Cainz's demand forecasting relied on experienced distributors and a fixed-order-quantity system using moving averages that could not accurately predict seasonal products or short-term sales trends. As the company planned to expand AI coverage across hundreds of stores, preprocessing data from even a handful of stores took three hours, making same-day completion at full scale impossible.

First attempt

Cainz initially used a third-party solution to train the demand model, but as the scope of demand prediction expanded, the solution showed its limitations and was replaced with Vertex AI Forecast.

Workflow diagram · grounded in source
1
Weekly sales data trigger
trigger
“the home improvement company uses sales data to train its demand prediction model once a week”
2
Parallel preprocessing with Cloud Run jobs
integration
“By using Cloud Run jobs, we were able to preprocess data in about 50 minutes, regardless of the number of stores”
3
Vertex AI Forecast model training
ai_action
“Vertex AI Forecast can handle large training data, is capable of multi-horizon predictions across multiple variables, and has a superior algorithm”
4
Explainable AI feature identification
ai_action
“Vertex AI Forecast has a feature called Vertex Explainable AI that streamlines this process and saves time by identifying the most effective features and explanatory variables for our system”
5
Results fed to core system
integration
“The results are then fed back into Cainz's core system”
6
Stock replenishment predictions output
output
“Successfully deployed the demand forecasting solution of Vertex AI across 209 stores, creating more accurate stock replenishment predictions”
Reported outcome

Cainz deployed its AI-powered demand forecasting solution across 209 stores, reducing data preprocessing time to 50 minutes regardless of the number of stores, and enabling more accurate stock replenishment predictions.

Reported metrics
Data preprocessing time50 minutes
stores with demand forecasting AI deployed209 stores
Previous preprocessing time baselinethree hours
Reported stack
Vertex AIVertex AI ForecastVertex Explainable AICloud Run jobsTAPBigQueryCloud StorageWorkflows
Source
https://cloud.google.com/customers/cainz
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Cainz deployed its AI-powered demand forecasting solution across 209 stores, reducing data preprocessing time to 50 minutes regardless of the number of stores, and enabling more accurate stock replenishment predictions.

What tools did this team use?

Vertex AI, Vertex AI Forecast, Vertex Explainable AI, Cloud Run jobs, TAP, BigQuery, Cloud Storage, Workflows.

What results were reported?

Data preprocessing time: 50 minutes; stores with demand forecasting AI deployed: 209 stores; Previous preprocessing time baseline: three hours (source-reported, not independently verified).

What failed first in this deployment?

Cainz initially used a third-party solution to train the demand model, but as the scope of demand prediction expanded, the solution showed its limitations and was replaced with Vertex AI Forecast.

How is this supply chain AI workflow structured?

Weekly sales data trigger → Parallel preprocessing with Cloud Run jobs → Vertex AI Forecast model training → Explainable AI feature identification → Results fed to core system → Stock replenishment predictions output.

Related supply chain cases