Guest Author Professor Pawar has two decades of industry experience in Supply Chain Management. Now an academic at SP Jain Institute of Management, he blends research with frontline conversations from leading practitioners. We invited him to share his perspective because it mirrors what we're seeing with our customers every day: the planning paradigm has shifted, and most organizations are still pretending it hasn't fundamentally changed.

President Dwight D. Eisenhower once reflected on his storied military career and observed something that sounds almost contradictory: "Plans are worthless, but planning is everything." Whether you're planning for a battle or something unexpected, nothing will happen the way you planned it. You'll throw your plans out the window. However, if you haven't been planning, you can't even start to work intelligently. The discipline isn't in the plan itself - it's in understanding the problem so deeply that when reality deviates, you know how to respond.

Supply chain planners live this every day. They're now facing unprecedented volatility, complexity that keeps growing, and a world that moves faster than any monthly planning cycle can handle. The question isn't whether your plan will survive contact with reality, it's whether your planning process prepares you to handle it when it doesn't.

The World Changed, and We're Still Planning Like It Didn't

For decades, we've had brilliant planning tools—MRP, APO, i2, JDA, Manugistics. They were built on a reasonable assumption: that the world is relatively stable and predictable. You forecast demand, optimize production, lock in a procurement plan, and execute.

This is a fatal assumption.

Think about what planners are dealing with now. Much has been written about the global events that disrupt supply chains on a regular basis. At the same time, companies offer ever increasing range of SKUs. Products launch and then become obsolete within months. Newer channels like quick commerce match hyper local demand and supply while promising delivery in 10 minutes. Customers are more informed and more demanding than ever.

In such an environment companies fight for a competitive advantage, however small, by solving more complex supply chain problems.

Take ride-sharing for example. What's an "SKU" when you're Uber? It's not just rides. It is rides from one neighborhood to another in a 30-minute window. You divide a city into one square-mile blocks, and suddenly you're forecasting demand for a matrix of thousands of origin-destination pairs of blocks, each with its own time-based pattern. On top of that, demand from one block another doesn't depend on its own history. It depends on how many people arrived at that block in the last few hours. You're forecasting a matrix based on a history of matrices, something that requires neural networks and leaves traditional planning methods behind.

Consider the case of quick commerce. Planners in this space are no longer just forecasting; they are running continuous 15-minute prediction-and-replenishment loops for specific SKUs to keep pace with hyper-local triggers. Demand here is no longer a historical trendline, it is a reaction to the present. A sudden rainstorm, a local sporting event, or a neighborhood traffic jam can instantly skew demand for specific items. You are no longer managing a static plan, you are managing a matrix of millions of micro-decisions that change every hour.

The bottom line? Traditional planning systems and processes are ill-equipped to handle the level of complexity inherent in modern business problems.

The Paradigm Has Shifted

Transforming for this reality requires three fundamental shifts:

1. Move from fixed forecasts to probabilistic forecasts
   Instead of saying "we'll sell 1000 refrigerators next month," start saying: there's a 50% chance we hit 1000, a 40% chance we reach 1,200, and a 10% chance we drop below 900. This changes S&OP from arguing about whether next month's forecast should be 1,000 or 1,050 to debating which scenarios to prepare for and which risks you're willing to take.

2. Move from Monthly Meetings to Always-On Planning
   Traditional S&OP is fixed against the calendar. By the time you close the month, run forecasts, collaborate on consensus, and hold meetings, you're halfway through the next month. Continuous planning replaces static planning cycles with a dynamic 'Sense-Think-Act' loop. An AI-enabled digital twin monitors real-time signals (from weather patterns to supplier risks) and perpetually runs thousands of 'what-if' scenarios to recommend optimized responses instantly.

3. Move towards flexible planning workflows.
   Traditional systems force all product categories, customer segments and geographies through the same planning workflow, across all time buckets. But different parts of your business need different approaches. So, the best planners build shadow spreadsheets because their planning tools fight them. Instead of rigid workflows, we need flexible platforms that empower planners to construct their own logic, tailoring workflows and mathematical models to fit unique business realities. 

What Kills AI Planning Projects

In many cases AI-driven planning is already delivering results at scale - UPS routes over 50,000 drivers daily with AI, saving millions in fuel costs. But plenty of such projects fail. Three reasons stand out:

1. Data Quality Kills More Projects Than Anything Else - If your data isn't standardized, clean, and aligned across sources, you're sunk. This is unglamorous work, but it's the difference between AI that works and AI that sits on a shelf.

2. Lack of Adoption Is the Silent Killer - Studies suggest 50-90% of AI projects fail to reach production. Teams run a pilot, which never gets scaled to production. This isn't a technology problem. It's a change management problem. AI implementation is a marathon, not a sprint, requiring time to calibrate models and stabilize workflows. Yet, the ROI is fast; organizations that commit to this calibration phase often see payback in under 12 months. 

3. Lack of Explainability Erodes Trust Fast - If the system can't explain why it's recommending something, planners won't trust it. Effective systems must function as a 'glass box,' presenting not just a recommendation, but a menu of ranked scenarios and the transparent logic behind every choice.

Why We Still Need Humans

Will AI replace planners? It will certainly replace the drudgery. As AI automates the transactional heavy lifting - spreadsheet wrangling and chart generation - planners are freed to become scenario strategists, cross functional collaborators and exception handlers.

However, raw processing power cannot replace human intuition. Zillow Offers serves as a cautionary tale: their AI-driven model crunched the numbers but missed the nuances, failing to account for factors like a neighbor’s noisy factory or the way sunlight hits a living room. The algorithm overpaid for bad inventory and underpriced good homes, ultimately forcing the division to shut down.

This illustrates the planner's permanent role as the 'Contextual Firewall.' Beyond just catching what the AI misses, humans must also enforce the ethical guardrails, ensuring that mathematical optimization doesn't override sustainability goals or corporate values.

What Leaders Need to Do Now

• Rethink Organizational Structure
  Siloed planning, procurement, transportation, and warehousing won't cut it anymore. You need cross-functional teams that blend IT, data science, and supply chain expertise.

• Invest in Change Management
  Research suggests you should allocate 20% of your budget to change management—process transitions, reskilling, stabilization time. If you're only budgeting for software, you're setting yourself up to fail.

• Redesign Skills and KPIs
  Stop rewarding firefighting and manual heroics. Start measuring decision quality, the number of scenario analyses run, and cross-functional collaboration.

• Shift the Culture
  Create a psychologically safe, experimentation-driven environment. Empower planners to work with and improve AI recommendations. That's how you build lasting, system-wide trust.

The Bottom Line

Eisenhower was right. Plans are useless. The moment you finalize a plan, reality changes. But planning, the discipline of thinking through scenarios, understanding probabilities, preparing your team to adapt, that's indispensable.

“The future of planning isn't about having the perfect AI solution. It's about having the best augmented humans; planners who can work alongside AI to navigate uncertainty, test scenarios, and make better decisions faster than the competition.”

The real advantage won't come from having the best AI. It will come from having people who know how to use it.