Using AI models to optimize best-selling inventory.

We created a new data-mart to consolidate multiple data sources, including ERP, point-of-sale, warehouse inventory, online sales, and loyalty program information, into a single data repository.

The outcome was a collaborative planning, budgeting, and forecasting solution with a more sophisticated algorithm for replenishment. This improved solution enabled them to interpret seasonality and extreme sales with a friendly user interface and a long-term view of their business performance. We used the IBM Cognos Analytics Framework metadata tool to load data-mart data into models that then could be manually extracted into Excel for specific suppliers.

By automating this task and other manual processes, we enabled our client to reduce human error and improve the efficiency of overhead operations. This series of custom solutions enabled us to develop a deep understanding of their infrastructure, business culture, and financial objectives. In addition, solution improved customer satisfaction. Our client asked us for input on how to optimize warehouse inventory to address issues such as seasonality, the economy, changing customer preferences, gross margins, and product availability. We recommended adding predictive analytics, Al, and machine learning to produce more precise predictive models and gain insights into how inventory selection and customer behaviour contributed to sales. We advised them to implement an Al and automation solution.

Our strategy focused on developing seven statistical Al models in using existing data to create more accurate inventory forecasts. Using Machine Learning, these results forecast conditions for up to 24 months into the future. We also created a link to facilitate inventory planning, budgeting, and forecasting as well as a link to advanced analytics reporting. We provided actionable insights into market basket analysis, which enabled decision-makers to filter and track promotions and returns.

By using basket analysis, we provide real-time insights into our client’s product value, which is helping the company better manage sales overall and increase margins while controlling inventory costs.

Temporarily closing the flagship store and the effect on store inventory.
KPI Digital immediately responded and changed the ETL to predict valid product inventory when the store reopens. We created a User Interface (UI) to define the store that we could reference to feed the Al forecast. This allowed for classes and subclasses of products to enter sales from different stores based on percentage of adjustments, multiplied by the rate of expected clients when the store reopens.