Using AI to optimize product variety: How companies can intelligently manage their portfolio

Modern companies face the challenge of managing their product or service portfolios in a way that maximizes customer value and profitability, while minimizing complexity and costs within the supply chain. A key factor in this equation is the breadth of the product portfolio. The larger and less structured the assortment, the greater the effort, cost, and complexity typically involved. Companies that don’t let their portfolios grow unchecked, but instead manage and streamline them actively, lay the foundation for efficient operations and avoid unnecessary burdens.

Portfolio management in the supply chain context means designing the assortment in a targeted manner: Which products and variants stay in the offering? Which are reduced or removed? The goal is to provide the right products in the right quantities to meet market demand without wasting resources. Decision-making is based on both market and demand data, as well as cost, production, and logistics information.

The Association for Supply Chain Management (ASCM) emphasizes that holistic product portfolio management is a crucial lever for building a more resilient and agile supply chain especially when new technologies such as artificial intelligence (AI) are applied strategically. AI can not only automate data processing but also support decision-making through pattern recognition and scenario analysis.[1]

Many companies still rely on traditional portfolio management methods such as manual assortment reviews, ABC analyses, or basic rules of thumb. But these approaches quickly hit their limits when product variety and market volatility increase. In practice, we often see an “oversupply” of products: too many variants in the system, many of which generate little to no revenue.

Years ago, a McKinsey study already pointed out the risks of having too few offerings for customers, while an oversupply of the “wrong” products leads to unnecessary costs and lost revenue opportunities. Traditional methods also tend to be retrospective, isolate product performance, and underestimate their full impact on the supply chain. Too many low-performing stock keeping units (SKUs) result in high inventory and setup costs, planning inefficiencies, and resource waste.[2]

Portfolio decisions are often made in silos – e.g., by product management or sales without input from production, logistics, or supply chain management. The result: poor responsiveness, inefficient inventory levels, and internal cannibalization effects within the portfolio.

• Centralized data platform
  All internal data from sales, procurement, production, and logistics, along with external signals (market trends, weather data, social media sentiment), feed into a data lake or cloud-based warehouse.
• AI analytics engine
  • Predictive analytics forecast SKU demand and flag potential bottlenecks.
  • Optimization algorithms balance service levels, inventory costs, and lead times.
  • Explainable AI adds transparency by identifying top drivers behind each recommendation (e.g., contribution margin or delivery risk).
• Scenario and simulation tools
  Interactive dashboards let users adjust parameters like demand trends or production capacity in real time and visualize the impact on inventory, costs, and service levels.
• Continuous feedback loop
  Productivity and sales data from the most recent period automatically flow back into the models, continuously refining AI forecasts and recommendations.[2]
• Seamless process integration
  Recommendations feed directly into Sales & Operations Planning (S&OP) processes and are automatically implemented through Enterprise Resource Planning (ERP) / Advanced Planning and Scheduling (APS) system interfaces.
• Governance and roles
  A clearly defined operating model assigns responsibilities (e.g., SCM, product management, finance) and ensures decisions are well-documented and traceable.

This target model combines agility, transparency, and sustainability and is key to continuously adapting the portfolio to dynamic market conditions while boosting efficiency and resilience across the supply chain.

• Reduced complexity and costs
  AI-powered portfolio analytics systematically uncover excess and inefficiencies. Studies show that targeted assortment streamlining can cut product costs by 10 – 20% and improve planning accuracy.[3]
• Faster response time
  AI systems automatically detect changes in demand or supply capabilities. This allows businesses to react more quickly to demand fluctuations, supply shortages, or market shifts. AI anticipates changes and helps synchronize supply with demand more effectively.
• Increased profitability
  By managing the portfolio based on contribution margins and strategic priorities, companies can improve their margins and foster innovation. Low-margin or non-differentiating products are identified and phased out more quickly.
• Enhanced resilience
  A focused portfolio reduces dependencies, simplifies phase-outs of underperforming SKUs, and strengthens the overall value chain.
• Real-world example: Unilever
  Unilever uses AI-powered portfolio optimization to streamline its product offering. A central tool in this effort is a data-driven platform that enables a holistic assessment of the portfolio through advanced analytics and insights. The tool evaluates whether a specific product should remain in the assortment or be discontinued based on its value to customers and Unilever itself. By removing underperforming items, Unilever creates space for growth, focuses more on core products, and develops new offerings aligned with evolving consumer needs.[4]

Deploying AI-powered portfolio optimization delivers clear value but also presents typical challenges. Here’s how we help companies navigate them:

• Data quality and availability
  Many companies struggle with fragmented or incomplete data that hinders decision-making. We help identify key data sources, assess quality and structure, and build a reliable foundation for analysis.
• Investment and business case
  AI projects need to make financial sense. We support the development of a business case for AI in portfolio management with clear use cases, realistic potential estimates, and measurable outcomes.
• Change management
  The shift to data-driven decision-making is as much cultural as it is technical. We support organizations in building competencies, clarifying roles, and driving adoption at all levels.

Our structured approach:

• Maturity assessment
  We evaluate the current data and system landscape as well as organizational readiness for data-driven decisions providing a realistic foundation for roadmaps.
• Target models & dashboards
  Together with clients, we define key KPIs, simulation logic, and visualizations to enable intuitive, transparent management.
• Technology selection & implementation
  We guide the introduction of suitable tools from flexible analytics platforms to explainable AI solutions that generate actionable and trustworthy recommendations.

Our goal is not to build a black box but a transparent, understandable, and connected decision-making tool that both business teams and management can rely on for smarter decisions with less complexity.

 

 

Sources:
[1] ASCM, 2025: “Product & Portfolio Management” [Aufgerufen: 9 Juni 2025]
[2] McKinsey & Company, 2020: “Mastering complexity with the consumer-first product portfolio”,  [Aufgerufen: 9 Juni 2025]
[3] World Economic Forum, 2025: “AI will protect global supply chains from the next major shock”, [Aufgerufen: 9 Juni 2025]
[4] Unilever, 2022: “Using AI to optimize our portfolio and fuel growth” [Aufgerufen: 9 Juni 2025]

In our series “AI in Supply Chain Planning”, we describe our top use cases of Artificial Intelligence in the supply chain. Previous use cases and other supply chain trends can be found in our insights.