Background
A global wine and spirits company owns a wide variety of beverage brands worldwide. With a consumer-focused approach, the organization strives to provide the most relevant products to each of their targeted customer profiles.
Challenge
The company began an engagement with Keyrus to scale their business intelligence (BI) team in order to deliver faster, newer reporting on projects. As a market leader, they knew that they needed to continue to invest and innovate to better understand their customers. This put pressure on the BI team to produce increasingly robust reporting and analytics.
The company found it difficult to hire or train employees for both analytical skills and beverage market expertise, so they decided to augment their staff with a strategic consulting partner who could scale with the demands of the team.
They were using Oracle BI as their main BI editor, but business users found it challenging, both to learn and to use. They wanted to move to a more self-service oriented tool to replace Oracle BI and move away from a traditional BI model. They needed a more scalable data visualization solution to reach their goals.
While implementing self-service analytics, the company simultaneously wanted to pursue advanced analytics in order to make measurable improvements, starting with account segmentation. They wanted to better focus field-sales efforts at the store level and help ensure that brands are positioned correctly during the sales process.
Solution
Self-Service Analytics
The organization had conducted an assessment of the leading self-service tools and chose Microsoft Power BI as a cost-effective editor. Keyrus then led a Power BI proof of concept (PoC) to assess the tool’s ability to address real-life business scenarios and recommend next steps.
The goals were:
Build a set of Power BI dashboards to address three use cases that were identified by the company to assess the tool’s ability to answer the business requirements
Guide business users on how to use Power BI for self-service analytics
Guide the IT team on how to govern Power BI in a bi-modal manner and assess the change impact on the organization
Recommend next steps to increase user adoption and successful roll out to the organization
After a successful PoC, our team designed and built a set of Power BI dashboards based on the specifications and mockups provided. The solution deployment and rollout included user acceptance testing (UAT), business and technical training, documentation, and solution handover to the IT team. The Power BI solution is a more modern, visual tool than the company’s previous solution.
Account Segmentation
The organization had started a strategic initiative on account segmentation to better focus field-sales efforts. With more than 250,000 active accounts in the US, they needed a low-touch automatic account segmentation process. They had conducted a PoC with the intent of using a supervised learning algorithm to learn how to automatically label accounts with proper segmentations.
The company engaged Keyrus to productionize the PoC and scale the main components of the machine learning system so account segmentation could proceed automatically with high confidence. We worked with their team to build decision trees in Python, perform data cleansing, and to generate outputs from various input data sources. Our team produced accuracy measurements to give a sense of confidence as to how well the machine was operating, and those metrics were shared all the way up to the executive level.
Our team’s knowledge in the beverage and alcohol space allowed for faster data model deployment and business user training.
Results
With a self-service BI model, the business no longer has to rely on the IT team to build new reports. Power BI dashboards provide live access to the data from any device and alleviate the burden of having to compile excel reports on a monthly basis. Tracking business performance with a self-service tool is now engaging and interactive.
The new Power BI dashboards allow the organization to:
Understand and analyze the performance of their products as distributors sell them to retailers. Managers can isolate markets/accounts/brands that are performing outside the norm and investigate what can be improved.
Understand what type of accounts the products perform well in and what future opportunities they may have, using an account segmentation matrix built with a machine learning model.
Take a deep dive into the account attributes of the retailers that sell the company’s brands. Analysts can understand their performance in high priority accounts and track their growth month-to-month.
Assist managers with understanding where the inefficiencies in their account management may exist; if an account moved from low priority to high, is it now selling enough?
Gain insight into the pricing strategy ROI and how to adjust for the upcoming months.
Evaluate brands in comparison with competitors within their division.
Analyze the performance of their brands in major transportation centers.
The outputs from the machine learning project are currently being deployed. They’ll be used as a measurement for the organization’s distribution partners in terms of target expectations for the upcoming fiscal year.
Read more about this project from the organization’s IT Director on our Clutch profile.
For more on Power BI, check out this blog post on how to perform concurrency testing of Power BI reports.