Impact of AI-driven demand forecasting on procurement efficiency in manufacturing firms

Project Abstract

AI-driven demand forecasting has emerged as a transformative lever for procurement efficiency in manufacturing firms by enhancing forecast accuracy, reducing stockouts, and lowering total cost of ownership across the supply chain. This study investigates how quantitative and machine learning-based forecasting models, including time-series, causal, and hybrid approaches, influence procurement decision-making processes, supplier collaboration, and inventory policy optimization in varied manufacturing contexts. Through a mixed-methods design, the research combines a quantitative analysis of historical procurement data from multiple manufacturing firms with qualitative insights gathered from procurement managers and suppliers. The quantitative component employs advanced forecasting algorithms (ARIMA, Prophet, LSTM, gradient boosting, and ensemble methods) to forecast demand at multiple levels (SKU, category, and plant) and compares performance against traditional baselines using metrics such as MAPE, RMSE, and sMAPE. The study links forecast performance to procurement outcomes, examining order quantities, lead times, safety stock levels, purchase price variance, supplier lead-time reliability, and fill rates. The qualitative strand explores decision-making rituals, data governance, change management, and the perceived barriers to adoption, including data quality, integration challenges, and organizational alignment. A key objective is to identify the mechanisms by which accurate demand signals translate into procurement efficiency gains, such as improved supplier negotiation leverage, reduced expedited shipping, and optimized lot sizing, while also revealing potential risks like model brittleness, forecast bias, and overreliance on automation. The results demonstrate that AI-driven forecasting significantly enhances procurement KPIs in firms with robust data infrastructure, cross-functional collaboration, and integrated ERP and supply chain planning platforms. In contexts with limited data maturity, hybrid models that incorporate expert judgment and external indicators (macroeconomic trends, supplier capacities, and seasonality) still deliver meaningful improvements, albeit with moderated gains. The research develops a framework for selecting forecasting models based on product heterogeneity, demand volatility, and supply risk, and presents a decision-support tool for procurement teams to calibrate inventory policies, safety stock, and order quantities in alignment with forecast confidence intervals. Sensitivity analyses reveal how forecasting horizon, service level targets, and supplier lead-time variability mediate the relationship between forecast accuracy and procurement efficiency. The study also assesses environmental and risk management implications, highlighting how precise demand signals minimize waste, reduce obsolescence, and strengthen resilience against supply disruptions. Practical implications include guidelines for data governance, model deployment architectures, and governance structures that facilitate ongoing monitoring, model maintenance, and continuous improvement. Theoretical contributions advance the understanding of information symmetry in procurement, the boundary conditions for AI adoption in operations, and the integration of machine learning with traditional inventory management theories. Overall, the findings affirm that AI-enhanced demand forecasting is a critical enabler of procurement efficiency in manufacturing firms, enabling more responsive procurement practices, improved supplier collaboration, and cost reductions while emphasizing the importance of data quality, governance, and human-in-the-loop oversight.

Project Overview

What This Project Is About

---

A plain-language overview of how manufacturers use forecasts to plan purchases, and how AI tools can improve accuracy and speed in predicting demand to guide procurement decisions.

---

---

The Problem It Addresses

---

Many manufacturers rely on simple guessing or historical trends to order materials. This can lead to stockouts or overstock, tying up money and slowing production. The project investigates how AI-based demand forecasting can reduce these risks and improve purchasing decisions.

---

---

Objectives of the Project

---

1. Explain how AI-driven forecasts differ from traditional methods.
2. Assess potential benefits for procurement efficiency and cost savings.
3. Identify challenges and data needs for implementation.
4. Provide a practical guide for small to medium manufacturers to adopt AI forecasting.

---

---

What You Will Do Step by Step

---

1) Review existing literature on demand forecasting and procurement. 2) Collect sample procurement data from a case study firm. 3) Apply simple AI forecasting tools to predict demand. 4) Compare AI forecasts with traditional methods and measure impact on stock levels and purchase costs. 5) Discuss practical implementation considerations. 6) Develop recommendations for practitioners.

---

---

Expected Outcome

Clear evidence on whether AI forecasting improves procurement efficiency, along with a practical implementation checklist and awareness of limitations for real-world use.