Case Studies

 

Over the course of the last few decades we have helped dozens of companies create data roadmap that convert data assets into innovative products @ scale. Below we offer a small snapshot of such cases.

Improving demand forecasting @ global food & beverage conglomerate.

PROBLEM:

  • Original forecasting solution in SAP was based on sales history (only) and average just above 60%

  • Can alternative data and machine learning improve forecast accuracy by 15% (leading to $50M+ in incremental sales)

SOLUTION:

  • Created data lake for Nielsen, POS, and weather data on Hadoop platform

  • Built a hierarchical machine learning model for demand

  • Designed report in PowerBI for ad hoc analysis by managers

OUTCOME:

  • Improve forecast across 3x3 product/market combination use cases by 20%

  • Created a roadmap for integration w/ SAP APO & TPM systems

Building a recommender system @ global steel producer.

PROBLEM:

  • Marketshare dropped during time of general industry growth, alerting customer to potential sales problem

  • Can recommender system help sales drive top-line growth by bringing the right product to the right customers?

SOLUTION:

  • Leveraged transaction data & customer demographics to create customer & product segments

  • Built recommendation engine to identify actionable insights for incremental sales in 10 weeks

OUTCOME:

  • ~1000K tons of incremental product and $100M+ sales captured for the top 10% of customers

  • Salesforce trained to use PowerBI dashboard & recommender system integrated w/ CRM

ant-rozetsky-SLIFI67jv5k-unsplash.jpg
DE57F129-A6C1-4B45-B302-B7EC84321B92.jpeg

Designing a platform & analytics center of excellence @ global entertainment & media conglomerate.

PROBLEM:

  • Develop a roadmap to build a platform and a center of excellence to transform into a customer-centric enterprise

  • Built an integrated ecosystem with enterprise data governance, model monitoring & development

SOLUTION:

  • Performed data management maturity assessment & remediation

  • Architected a data governance warehousing & governance framework w/ master data management system

  • Integrated w/ AWS systems for BI and dash boarding

  • Designed Data as a Managed Service platform

OUTCOME:

  • Improve service by providing personalization to users

  • Establish golden view of customer & increase CLTV

Creating a servitization platform @ global tire manufacturer.

PROBLEM:

  • The earthmover division was rapidly losing market share to competitors from China due to commoditization

  • Can a service offering be developed to enhance product for mining corps where tires make up to 10% of operating cost

SOLUTION:

  • Suggested recommendation engine to create value-added insights from existing data for 200+ mining clients worldwide

  • Performed exploratory data analysis to understand tire wear as a function of 500+ variables from dozen sources

  • Built predictive model for failure rates relative to operating parameters in R on Hadoop w/ 7% margin of error (2σ)

OUTCOME:

  • Created a service division and a new business model with $50M a year in incremental sales and 20% retention boost

jahongir-ismoilov-AffCPyYCZUQ-unsplash.jpg
nasa-_SFJhRPzJHs-unsplash.jpg

Building an anomaly detection system @ large US utilities operator.

PROBLEM:

  • Vendors and contractors have defrauded the operators of millions through complex transactional schemes

  • Can a system be developed to proactively identify potentially fraudulent transactions?

SOLUTION:

  • Leveraged unsupervised learning clustering algorithms (isolation forest) to group “similar” sets of transactions

  • Developed an interactive Tableau dashboard for review of “flagged” cases & daily monitoring

OUTCOME:

  • Robust monitoring system that flags suspicious activity and enable proactive monitoring of vendors and contractors

Increasing sales @ global talent & media agency.

PROBLEM:

  • Can the firm drive growth for itself and increase value to clients by uncovering authentic brand/client relationships?

  • Will machine learning techniques applied to 3rd party social media data be used to equip agents with innovative insights?

SOLUTION:

  • Collected data from dozens of curated social media sources

  • Used machine learning to create affinity and influence scores and construct a knowledge graph b/w talent & brands

  • Created an interactive dashboard w/ talent-based and brand-based rankings, trained half a dozen agents

OUTCOME:

  • Increase value of sponsorships for brands by increasing market penetration & for talent via leverage in negotiations

  • “Stole” key A-list clients from competitors by running their names through graph and creating compelling brand offers

pinho-yjXAtMCPdGs-unsplash.jpg