Enhancing Operational Process Performance with a Modern Cloud Platform

Background

A notable company in the network and app intelligence space, which Keyrus provided a cloud-based portfolio of pre-packaged use cases for to help its customers analyze, optimize, and monetize application experiences using machine learning-based insights and real-time actions. By working with the largest service providers around the world, the company serves billions of individual users across over 100 different countries.

Challenge

Following the successful modernization of the company’s financial analytics, the purpose of this project was to extend the modernization to further enhance their manufacturing analytics capabilities through addressing various key areas: 1) Data integration of 20+ new tables from recently established data pipelines for SAP & Salesforce, as well as a net new flat file SFTP source. 2) High Data Refresh Frequency was required for one of the reports (<15mins from source to dashboard) - Control of access & permissions in the final analytics solution to separate by department & role - Improving data management practices, including data quality and governance. - Enhancing operational efficiency by providing automated, meaningful, and visually engaging data analysis to plant managers, manufacturing analysts, and executive leadership. - Eliminating reliance on outdated reporting by empowering stakeholders with reliable & highly available insights that are accessible via interactive & engaging dashboards.

Approach

To correctly plan for and execute on the many aspects of this project, the 20 weeks of work was broken up into 2 phases (4 weeks of Discovery & 16 weeks of Implementation), with the following detailed approach: Discovery Given the scope of the project, a discovery sets the stage for an efficient implementation by following these keys steps: • In-depth requirements gathering to fully understand what is expected in the final data solution • Review & analysis of existing reports & data structures to identify what can be leveraged to support the new data platform and what gaps need to be addressed to fully deliver a successful solution • High-level architecture design to broadly map out the different layers & relationships within the final manufacturing data warehouse • End-to-end project plan outlining an iterative development cycle (design, build, test, and deploy) that orchestrates all the key delivery dates & relevant pre-requisites during the implementation phase Implementation • The implementation builds on the initial discovery by applying the designated plan and progressing through the core project work in an agile fashion: • Data Ingestion relied on Fivetran (cloud sources) and HVR (local network sources) to connect to the various SAP, Salesforce, and flat file data through daily synchronizations into an initial landing environment within Snowflake. • Datawarehouse ELT brough the data through various layers of transformations done by DBT before arriving at a centralized reporting schema also within Snowflake. • Data visualization is made accessible for consumption through Tableau, an industry leading data visualization software. This enables executives, manufacturing analysts, and plant managers to explore and visualize various manufacturing analysis. • Security is handled in 2 layers: o Row & column level restrictions applied at the workbook & metric level. o Departmental segregation via permission groups and projects (folders). • Governance is established through automatic data quality checks controlled in DBT. Processes for new content deployment (in the DWH and/or on Tableau Cloud) & viewer/explorer access requests maintain the right checks & balances for proper data development & consumption. Team • Daniel Vaucrosson, Managing Consultant • Leslie Potgieter, Data Architect • Yi-Jen Tu, ELT Data Consultant • Luke McCully, Visual Analytics Data Consultant By implementing this comprehensive approach, a robust pipeline and process for Manufacturing analysis has been established, enabling efficient data ingestion, computation, storage, and consumption of results for key stakeholders.

Key results

Benefits

• Automated Ingestion & Consolidation of Multiple Sources into a single data warehouse that can be used for many current & future analytics applications. • Up to 24-hour reduction in refresh frequency for operational reports (now <15mins). • Interactive Multi-Source Visual Insights, previously not possible or done manually with large effort and delayed action. • 1-5 days saved per set of ad-hoc analysis now done almost instantaneously. • Reliable & Secure Data, correct access for all relevant users and automated quality checks & alerts to ensure data can be trusted. • Near real-time identification & resolution times for data issues vs previously taking up to 14 days reduction.

Share this key play