_edited.png){kind=tracking}
Phase 1: Data Engineering
GOAL: Company desired to limit resources needed to troubleshoot daily performance issues, deliver future reporting capabilities, resolve the high feeling of inconsistent data across the company, and minimize weekend work needed to ensure data systems were in sync. Their existing environment included an old SQL server based on-prem, databases with 5,000+ tables, tons of mapping and bridging tables, legacy Bl tools, and developer dependent reporting systems.
INSIGHT + ACTION: Beacon first worked closely with key stakeholders to architect, organize, engineer and cleanse company data. We removed all reporting (Ad Hoc, SSRS, Automated) from the production servers to eliminate the security and performance issues linked to them directly accessing production databases. Implementing a cloud data lake/ data warehouse and no-code integration provided better scalability, a single source of truth for all reporting, and an ability to easily add new data points. After identifying 300 tables that were important, we consolidated them to under 40 tables exposed to end users. Finally, a self-service Bl platform was deployed and optimized for their presentation layer empowering non-technical users and freeing up IT team resources.
TECH STACK: Data lake and data warehouse hosted on AWS (Redshift , S3). Implemented Hevo for data pipeline as a service and DBT and Airflow for data wrangling. Power BI deployed as their Bl platform.
BUSINESS OUTCOME: Removed security and performance issues, restored faith in data quality, removed weekend system update work, and laid the foundation for machine learning.
Increasing Inventory Turnover.
Phase 2: Data Science
GOAL: The company sought innovative ways to use it’s data to accomplish 3 initiatives: better forecast supply and demand, optimize freight routing and timing, and reduce lost inventory. Accomplishing these objectives would increase sales, decrease idle inventory, reduce short lead and overall freight costs, and save on capital expenditures.
INSIGHT + ACTION: Beacon first worked with stakeholders and key team members to organize, engineer and cleanse company data. Next, we developed advanced customer profiles and variables to feed into machine learning models. The models predicted inbound inventory and availability down to the SKU level and by company location. Additional models predicted ideal freight routes and optimal final destinations on a load-by-load basis, which could be used to avoid intermediate stops and last minute shipping at expedited rates. All these tools were built and delivered to the team in custom dynamic BI tools and reporting for each business function.
BUSINESS OUTCOME: This project is expected to decrease idle inventory up to 40% and improve inventory forecasting accuracy by over 30%.
“I have worked in the consulting and solution delivery space for a while, and just the feeling my team and I received from the start was that, it felt as if Beacon was another department here at Tosca. Very personable and the inclusion of our core Subject Matter Experts from the start was great.”
“The amount of and types of resources we had access too at any point is what has surprised me most working with Beacon. I really appreciate the fact that at many points of the project, they were willing to jump on a plane to work through any hurdles onsite. From the Founding Partner to the Data Engineer, everyone is in the loop, which makes things pretty seamless.”
- Jeff Etringer, Tosca Integrations & Development Manager