How can we get our demand forecast accuracy to 90%?

This was the question thrown at us during a recent customer meeting. While the customer is a B2B organization this challenge is universal. 10s of product groups, 100+ products, and probably 1000s of SKUs. Multiple manufacturing locations, delivery timelines, and various delivery locations. Existing orders, current inventory, production capacity, and raw material lead times. Throw in seasonality, geo variability, organizational priority into the mix and you have the most organizations struggling to get to 70% mark. So this organization definitely wanted to raise the bar.

And even if one is not thinking of getting that far, every organization wants to improve their accuracy. The benefits are obvious – better alignment of demand and supply leading to lower supply costs, higher revenue on account of no opportunities left on the table, optimum inventory and production lines. And most importantly a satisfied customer!

Here are five suggestions to enhance your demand planning process:

1. Collaborate Across Departments:

Effective demand planning involves cross-functional collaboration. Engage sales, manufacturing, finance, supply chain, and other relevant teams to share insights and align on demand forecasts. A collaborative approach ensures that the demand plan reflects inputs from various perspectives and results in a more accurate prediction. Easier said than done, but not optional. So get across those departmental boundaries and rise above the organizational politics and leverage the collaborative wisdom.

2. Bring multiple scenarios into the consensus meeting: Its likely that every function will have their own priorities and therefore their respective preferences. To showcase the real picture it always helps to draw them multiple scenarios – aggressive (sales), conservative (production), or realistic (planning). This gives voice to everyone’s opinions and quickly illustrates the direction towards the consensus plan.

3. Utilize Accurate Data and Advanced Analytics:

This is probably the biggest struggle and exhaustion which planners face. Availability of all (a) the relevant data (b) in the required format including (c) historical (previous forecast vs actuals) and (d) transactional data. Due to this they end up spending 65-70% of their time every cycle in getting the pieces together and only 30% time in analysis, brainstorming and developing the final plan. The ease of technology adoption has increased so much that there is no reason to hide behind spreadsheets.

One doesn’t need to be a rocket scientist to be able to deploy predictive or AI-based solutions for demand prediction. Begin by collating all the relevant historical sales data, customer behavior, market trends, and other external factors that impact demand. Use advanced analytics tools and machine learning algorithms (linear regression, exponential smoothing, multiplicative decomposition, CNN-QR, DeepAR+) to identify patterns and forecast demand more accurately.

4. Implement Demand Sensing Techniques:

Demand sensing techniques leverage real-time data, such as point-of-sale data, social media trends, and weather patterns, to adjust forecasts dynamically. By incorporating this dynamic data into your demand plan, you can respond promptly to unexpected changes in customer behavior and avoid inventory shortages or excesses. Incorporating real-time data can also help you respond quickly to changing market conditions.

5. Continuously Monitor and Refine the Plan:

Demand planning is an iterative process. Regularly review and compare actual demand with forecasted values to identify discrepancies and improve forecasting accuracy. Collect feedback from various stakeholders and update the demand plan as new information becomes available. Continuous improvement ensures that your demand plan remains relevant and effective in a dynamic business environment.

Improving your demand plan can collectively impact multiple different KPIs - inventories, capacities, customer sat, of course cash flows. So its bit like killing multiple birds with one stone. Bring your collective knowledge and the most relevant technology to the task. Nothing can be more important.