Mark Anthony Group: How GenAI Improved the Synergy Between Supply, Demand, and Marketing

Refreshing the Formula: Tackling Consumer Demands with Data and AI in the Beverage Industry 

Mark Anthony Group. a leading beverage manufacturer, recognized the need to keep up with intensifying competition in a rapidly evolving market of consumer preferences. They aimed to significantly improve marketing and sales insights across the organization, enabling them to deliver unprecedented value to stakeholders. Each function, including demand, supply chain, marketing sales and finance, relied on its own reports and spreadsheets, leading to conflicting "versions of truth" and lengthy reconciliation cycles.  

Their current system was complex and fit for purpose. Users needed to heavily rely on IT to run reports, making it less than ideal for analytics and curated solutions. They needed a way to make their system scalable, implement proper architecture and governance, and establish self-serve analytics. Analysts and business users spent most of their time extracting, reconciling, and preparing data instead of analyzing it or experimenting with new ideas, which slowed down decision-making and limited innovation. 

It became clear that they first needed to modernize their foundational data platform to implement generative AI for improved insights. However, recognizing the need for change and implementing it are two different challenges. Our client understood that achieving deeper, actionable insights required more than just advanced technology; it necessitated strategic leadership. They needed a clear, visionary roadmap on how to leverage their new capabilities to not only meet but exceed the expectations of their stakeholders. They also needed a way to safely experiment with GenAI, proving value on targeted use cases before rolling it out widely across brands, markets and lines of business.  

Reimagining Analytics: Building an Agile, AI-Enabled Data Platform 

As an AWS Advanced Tier Services Partner with deep expertise in cloud analytics and a comprehensive arsenal of AWS solutions, Adastra became the driving force behind MAG's digital revolution. Together, we designed a modern data and AI operating model that combined a centralized engineering "factory" with business-embedded sponsors in demand, supply chain, marketing, sales, and finance, ensuring every use case was tied to concrete business outcomes. 

Building a Data Lakehouse 

To enhance data access and query response capabilities, we embraced a cutting-edge lakehouse architecture, anchored by the robust and scalable Amazon S3. This foundational shift was augmented by leveraging AWS Glue and Lambda, along with the powerful querying capabilities of Amazon Athena, creating a seamless and integrated data ecosystem. On top of this, we introduced a refined data layer with business-friendly models that could be accessed directly for self-serve analytics and also feed advanced ML and GenAI workflows. 

This strategic architectural framework not only prepared the ground for sophisticated analytics but also ensured smooth integration with generative AI tools. As Elkin Arboleda, AWS Practice Lead at Adastra, aptly noted, "AWS services integrate easily with each other with very little coding needed to use them." This synergy between AWS solutions not only streamlined the integration process but also significantly accelerated the pace of our digital transformation. By decoupling storage from compute and standardizing integration patterns, MAG can now add new data sources and use cases without re-architecting their core platform. 

Cleaning the Data  

Aiming to unlock the vast potential of its data assets, we implemented rigorous data quality management practices, where data undergoes quality checks in the cleansed layer and is then transformed into the main "refined" layer of the data lake. The refined layer adds a data model on top of most of the data. From there,  data can be consumed, pushed downstream, or exposed through APIs.  

AWS services were used for data processing, governance, and storage, using Amazon S3, AWS Lambda, AWS Glue, AWS Step Functions State Machines, Amazon SQS's, Amazon EventBridge, Amazon EC2 and Amazon RDS. All processes are fully auditable, from start to finish, across components and the audit information is available through Amazon S3 directly or Amazon Athena, or through a special-purpose software development kits (SDKs). This end-to-end auditability not only supports regulatory and privacy requirements, it also provides the traceability needed for responsible AI, since every GenAI answer can be tied back to governed data sources and transformations. 

Deploying Advanced Analytics 

"When we rolled it out, it was 100% adoption. Within the first month, every sales rep was using the AI Sales Activity Recommender that we built."

– Sam Wong, Senior Director of Data, Analytics and AI

Once cleansed, the data can be used as input for advanced analytics and machine learning models.  We deliberately started with a GenAI sales assistant because sales is closest to revenue, making it the fastest path to demonstrating clear ROI and building organizational buy‑in for AI. We deployed Amazon SageMaker and Amazon Bedrock to develop an AI sales assistant and marketing "Prescriptive Sales Recommender" sentiment tool. The GenAI sales assistant allows commercial teams to ask natural-language questions such as "What were last quarter's depletion trends for White Claw by channel?" and receive instant, explainable answers grounded in curated data. The prescriptive recommender combines sentiment, historical sales, and promotion data to suggest which brands and SKUs to prioritize by market, channel, and timing, directly informing campaign planning and trade spend decisions. 

Having a strong data foundation ensures the group is future-ready, with the requisite data and machine learning pipelines to make the analytics process more efficient. This will make it easier to innovate moving forward and integrate with additional AWS services. New GenAI use cases, such as demand-planning copilots or financial scenario simulators, can now be prototyped quickly on top of the same trusted data, without rebuilding pipelines.  

Driving Success with AI: Boosting Efficiency, Satisfaction and Industry Leadership 

The transformative journey has led to a remarkable enhancement in productivity, driven by deeper insights into their operations. This shift has had a profound impact on the work culture, fostering increased job satisfaction among team members. Freed from the confines of monotonous, labor-intensive tasks, the team has experienced an increase in purpose and efficiency. From the outset, leadership positioned GenAI as "augmented intelligence," a way to supercharge people's capabilities rather than replace them, which drove strong engagement and adoption across functions. 

The project allows the client's demand planning team to accurately report on consumption/depletion forecasts, financial planning forecasts, and agency supplier financial planning forecasts to proactively manage business processes and stock. Demand planning, supply chain, marketing, sales, and finance now align on a single, governed view of the truth, improving service levels while reducing costly overstock or stockouts. 

The engagement also resulted in:  

• A single source of truth for cleansed, governed data used consistently across demand, supply chain, marketing, sales, and finance, rather than fragmented reports maintained by individual data providers 
• 100% adoption of the GenAI sales assistant by the commercial team, driven by immediate, tangible value in answering day-to-day sales questions 
• The democratization of descriptive analytics, visualization, and data insights through self-serve access to the refined data layer and AI sales assistant, significantly reducing dependence on IT for routine reporting 
• Faster, more collaborative planning cycles, with all teams working from the same metrics and assumptions 
• Actionable insights on marketing and sales processes for data-driven decision-making, including which promotions, channels, and messages are most effective by market 
• GenAI use cases can be rapidly deployed 
• Cost-effective innovation  
• Early pilots have shown substantial time savings in reporting cycles and improved accuracy of key forecasts, and the flexible AWS lakehouse architecture ensures that the solution can seamlessly scale, unlocking unprecedented insights and innovation, and positioning the client for sustained leadership in a competitive landscape