01 — Overview / Context
What was the problem we were solving?
B2B supply chains are fast-moving, interconnected systems operating at massive scale. Demand planning means forecasting across millions of SKUs every month, and minor errors cascade quickly into empty shelves, broken supplier relationships, and lost revenue. Forecasting demand plans across hundreds of products is complex, and minor errors quickly lead to empty shelves. Our mandate was to understand where the planning process was breaking down, and design systems that meaningfully improved both decision-making speed and quality.
Key Stakeholders: Demand Planners, Prediction Models, and Inventory Management
Demand plans include millions of SKUs forecasted every month or planning cycle
Supply decisions inform stakeholders, retailers, and global branches across cycles that take weeks
02 — Research / Barriers
What did we learn from users?
We used a systems design approach, conducting deep qualitative and contextual research to understand how decisions were actually being made, and precisely where they broke down.
14+
Interviews
100+
Minutes of observations
3+
Workshops and roundtables
3
Contextual inquiries
Key Barriers Identified
Cognitive Overload + Task Fatigue
Demand Planners spend ~90% of their time on repetitive tasks with limited context, leaving almost no room for strategic thinking or high-quality decision-making.
Experience Know-How + Process Inefficiency
Planners edit demand plans relying on tacit knowledge that is hard to formalize, and resist handing off decision-making control, because the system can't reflect what they know.
Risk Aversion + Overcorrection
9/10 planners say they have to defend every update and change, but have no structured way to capture or show their logic, creating constant credibility pressure.
Availability Bias + Perceived Pattern Recognition
Planners struggle to trust statistical model recommendations they don't understand, and regularly adjust or reject forecasts based on instinct over evidence.
03 — Reframe / How Might We
Demand planning becomes a cycle of guesswork and chair-swiveling across tools, driven by rushed timelines and limited context, despite access to advanced enterprise systems. This creates chaos and longer planning cycles in environments where speed is everything.
How Might We
How might we reduce uncertainty and activate clarity, collaboration, and confident decision-making during critical demand planning moments across all stakeholders?
/ New Paradigm / Solution Themes
Shift demand planning from reactive editing to focused & intentional decision-making using explainability, traceability, controls and proactive system learning.
/ Strategic Values For Tool /
Pillar #1
Improve signal quality and reduce noise
with high-stakes decisions, while preserving agency.
Pillar #2
Make decision logic transparent
and reduce repetition, to build confidence and clarity.
Pillar #3
Embed continuous learning
and improvement mechanisms across the system, with autonomy.
Pillar #4
Enable cohesive collaboration for defensible actions
at critical moments.
04 — Interventions
What did we design to address this?
05 — Impact
What changed as a result?
70+
Screens designed across an enterprise-grade system for 3+ Clients
Across 7 core product areas, the system delivered a scalable decision architecture with out-of-the-box capabilities, prototype secured multi-client buy-in.
$2.2M+
Revenue Unlocked From MVP
A defensible decision system that supported over $2M in pilot forecasted revenue and directly strengthened enterprise sales momentum.
3–5×
Design Velocity Scaled For Launch
Faster production enabled by embedding structured service-design informed system thinking through reusable, AI-assisted design system scaling.
06 — Learnings
What would we do differently?
Learning 01
Prioritizing Progress Through Iteration
Complex systems require staged solutions. Break problems down and move forward in deliberate, sequenced iterations, trying to solve everything at once stalls momentum and obscures signal.
Learning 02
Balance Autonomy & Adoption With Control
Users want autonomy, yet need support. Adoption depends on simultaneously addressing trust, fear of replacement, and education, you can't design for one without the others.
Learning 03
Choosing Artifacts That Drive Decisions
Not all artifacts translate into action. The right artifacts must be chosen to support decision-making beyond thinking, outputs that inform without enabling action aren't enough.