Inventory Demand Forecasting

Why is supply chain still struggling with demand forecasting? Maybe because we try too hard to explain demand instead of recognizing its context. We spend years modeling price, promotions, seasonality, macro, weather, trying to explain demand. But markets behave less like physics and more like human systems: adaptive, emotional, nonlinear. David Epstein describes a useful shift in Range. Netflix stopped trying to decode what makes a movie good. Instead, they asked: who is this user similar to, and what did they like? Analogy replaced explanation. This technique isn’t unique to Netflix. - Medicine predicts outcomes using case-based reasoning / patient similarity analytics. - Climate science uses analog forecasting. - Banks estimate risk through peer group and cohort models. In all these domains, similarity-based inference outperforms causal explanation when systems are complex and adaptive. So what if we flipped demand planning the same way? Instead of asking: “Why will this product sell?” Ask: “Which past situations looked like this and what happened next?” For example, instead of forecasting SKU 123, define the situation: FMCG staple, low price, GT-heavy channel, low promo, high inflation, festival season, rising volatility. Then find similar past situations and observe what happened next. So instead of saying: “Demand will be 12,340 units.” You say: “In 37 similar situations, average uplift was +9%, with a 70% chance it will be between +5% and +14%.” Not predicting demand. Recalling it from history’s closest analogs. This gives planners not just a forecast, but also confidence and risk. I’m looking for a few volunteers to test this approach in practice, reach out if you’d like to explore. #SupplyChain #DemandForecasting #Analytics #AI #MachineLearning #SystemsThinking #DecisionScience

Inflation isn't just about rising prices; it's a catalyst for changing consumer behaviors. As purchasing power shifts, businesses must adapt swiftly to meet evolving demands. Hindustan Unilever Limited (HUL), a leader in the FMCG sector, showcases how embracing AI can turn these challenges into opportunities. 📌 The Challenge #HUL observed significant fluctuations in demand across its diverse product portfolio during inflationary periods. Premium products experienced slower sales, leading to overstock situations, while budget-friendly items frequently faced stockouts. Traditional forecasting methods, relying heavily on historical sales data, struggled to keep pace with these rapid changes in consumer preferences. 📊 The Solution: AI-Driven Demand Forecasting To address this, HUL integrated AI-powered analytics into its demand forecasting processes. This advanced system enabled the company to: Analyze Real-Time Consumer Behavior: By examining current purchasing patterns and consumer sentiment, HUL could detect emerging trends and shifts in preferences. Incorporate External Economic Indicators: The AI model factored in various economic indicators, such as inflation rates and consumer confidence indices, to predict their impact on product demand. Optimize Inventory Management: With precise demand forecasts, HUL adjusted its inventory levels accordingly, ensuring optimal stock across all product categories. 🔹 Key Insight: The AI-driven approach revealed that demand for budget-friendly products was increasing at a rate three times higher than traditional models had predicted, while premium product sales were declining in specific regions. 📈 The Impact 20% Reduction in Unsold Premium Stock: By aligning inventory with actual demand, HUL minimized excess stock of premium items. 35% Improvement in Stock Availability for Budget-Friendly Products: Ensuring that high-demand, cost-effective products were readily available led to increased customer satisfaction. Enhanced Revenue and Profit Margins: Optimized inventory management reduced holding costs and prevented lost sales, positively impacting the bottom line. 💡 The Lesson In times of economic uncertainty, relying solely on historical data can be a pitfall. HUL's proactive adoption of AI-driven demand forecasting exemplifies how leveraging advanced analytics allows businesses to stay agile and responsive to market dynamics, ensuring they meet consumer needs effectively How is your organization utilizing data analytics to navigate market fluctuations? #datadrivendecisionmaking #businessstrategies #dataanalytics #demandforecasting

🍦 𝗔𝗜 𝗖𝗮𝘀𝗲: 𝗨𝗻𝗶𝗹𝗲𝘃𝗲𝗿 𝗜𝗰𝗲 𝗖𝗿𝗲𝗮𝗺 — 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 𝗧𝗵𝗮𝘁 𝗥𝗲𝗮𝗰𝘁𝘀 𝘁𝗼 𝗪𝗲𝗮𝘁𝗵𝗲𝗿 & 𝗦𝘁𝗼𝗿𝗲 𝗥𝗲𝗮𝗹𝗶𝘁𝘆 🤔 AI in supply chains isn’t just a promise — it’s already delivering measurable results. 🌡️ 𝗨𝗻𝗶𝗹𝗲𝘃𝗲𝗿’𝘀 𝗘𝘂𝗿𝗼𝗽𝗲𝗮𝗻 𝗶𝗰𝗲 𝗰𝗿𝗲𝗮𝗺 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 faces rapid, weather-driven demand swings. Seasonal volatility often outpaces traditional forecasts, leading to lost sales and waste. 📣 𝗛𝗼𝘄 𝗔𝗜 𝗵𝗲𝗹𝗽𝗲𝗱 𝗨𝗻𝗶𝗹𝗲𝘃𝗲𝗿’𝘀 𝗗𝗲𝗺𝗮𝗻𝗱 𝗳𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 & 𝗱𝗲𝗺𝗮𝗻𝗱 𝘀𝗲𝗻𝘀𝗶𝗻𝗴 ▪️ Uses daily weather updates from hyperlocal data (temperature, rainfall by city). ▪️ Pulls live data from AI-enabled freezers with IoT sensors tracking SKU presence and quantities. ▪️ Combines POS and distributor sales to reconcile forecasts in near-real-time. ▪️ Adds event and promotion data to refine demand signals. 𝗧𝗵𝗲 𝘀𝘆𝘀𝘁𝗲𝗺 𝘂𝘀𝗲𝘀 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗳𝗼𝗿 𝘀𝗵𝗼𝗿𝘁-𝘁𝗲𝗿𝗺 𝗱𝗲𝗺𝗮𝗻𝗱 𝘀𝗲𝗻𝘀𝗶𝗻𝗴 𝘁𝗼 𝗱𝗲𝗹𝗶𝘃𝗲𝗿: 🔹 Weekly rolling forecasts that adjust monthly plans. 🔹 Daily alerts so teams can replenish high-demand SKUs fast (e.g., +5°C triggers orders within 48 hrs). 🔹 Inventory reallocation from low- to high-demand areas before expiry. 📈 𝗞𝗲𝘆 𝗥𝗲𝘀𝘂𝗹𝘁𝘀: ✔️ 10% higher forecast accuracy, reducing waste and missed sales. ✔️ 30% higher retail orders due to proactive replenishment and SKU mix optimisation. ✔️ Lower waste through stock reallocation in cooler periods. ✔️ Faster decisions — from a week to hours. 📍 𝗧𝗵𝗶𝘀 𝘀𝗵𝗼𝘄𝘀 𝗵𝗼𝘄 𝗔𝗜 𝗰𝗮𝗻 𝘁𝘂𝗿𝗻 𝘄𝗲𝗮𝘁𝗵𝗲𝗿 𝗮𝗻𝗱 𝘀𝗮𝗹𝗲𝘀 𝗱𝗮𝘁𝗮 𝗶𝗻𝘁𝗼 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗳𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝘀 𝘁𝗵𝗮𝘁 𝗰𝘂𝘁 𝘄𝗮𝘀𝘁𝗲, 𝗯𝗼𝗼𝘀𝘁 𝘀𝗮𝗹𝗲𝘀, 𝗮𝗻𝗱 𝘀𝗽𝗲𝗲𝗱 𝘂𝗽 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗲. 👇 𝘞𝘩𝘢𝘵 is 𝘩𝘰𝘭𝘥𝘪𝘯𝘨 𝘭𝘰𝘤𝘢𝘭 𝘤𝘰𝘮𝘱𝘢𝘯𝘪𝘦𝘴 𝘧𝘳𝘰𝘮 𝘭𝘦𝘷𝘦𝘳𝘢𝘨𝘪𝘯𝘨 𝘈𝘐 𝘪𝘯 𝘴𝘶𝘱𝘱𝘭𝘺 𝘤𝘩𝘢𝘪𝘯𝘴?

How Zinnia Used AI to Forecast Daily Call Volumes with 95% Accuracy 📞 At Zinnia, we needed a better way to forecast call center volumes — our existing tool often missed the mark by 10–20%, making staffing plans unreliable So, we rolled up our sleeves and built our own AI forecasting solution: ✅ Combined Prophet (seasonality + trends) with XGBoost (learn from errors) ✅ Used real-world signals like holidays, month-ends, and even Mondays after weekends ✅ Tuned everything with time-aware cross-validation We tested A LOT (even LSTMs and SHAP-based pruning!), but our hybrid model consistently delivered 95%+ accuracy across clients. 🔍 I’ve shared the full breakdown, code, and what worked (and what didn’t) in this Medium article — practical, real-world AI for ops. If you're a data scientist, ML engineer, or even an ops leader — this one’s for you. Josh Everett | Pawan Choudhary | Daniel Gremmell | Eti Gupta #DataScience #Forecasting #AI #XGBoost #Prophet #TimeSeries #MLinProduction #CallCenterAI #WorkforcePlanning #ZinniaTech #AIinOps

Ever wonder why some e-commerce brands always seem to have the right products in stock, while others struggle with overstock or empty shelves? It all comes down to demand forecasting—and in 2025, it’s getting an AI-powered upgrade. ● From guesswork to precision Traditional forecasting relies on historical sales data. AI-driven tools now go beyond that, integrating real-time factors like weather, local events, and even social media trends. The result? Forecasts with 90%+ accuracy instead of the usual 50%. ● GenAI: the next step Generative AI takes it further by analyzing unstructured data (customer reviews, trends, emerging demand signals) and answering questions in plain language. No more complex spreadsheets—just instant insights for better inventory planning. ● AI tools leading the way: ✔ Simporter – AI-powered forecasting that integrates multiple data sources to predict sales trends. ✔ Forts – uses AI for demand and supply planning, ensuring optimized inventory. ✔ ThirdEye Data – AI-driven forecasting that factors in seasonality and customer behavior. ✔ Swap – AI-based logistics platform that enhances inventory management. ✔ Nosto – AI-driven personalization that recommends the right products at the right time. ● Why this matters for #ecommerce? ✔️ Avoid stockouts that frustrate customers ✔️ Reduce excess inventory and free up cash ✔️ Adapt quickly to market shifts How are you managing demand forecasting in your store? #shopify

Amazon Data Science Interview Question: Your model predicts demand for products, but a new product has no historical data. How would you handle the cold-start problem? Before jumping into a solution, I'd clarify two things. First, what's the business objective—is this forecast being used for inventory planning, replenishment, or procurement? Second, what's more costly: over-forecasting or under-forecasting? That helps determine both the modeling approach and the evaluation metric. Assuming this is for inventory planning, the goal isn't simply to minimize forecast error. It's to reduce stockouts while avoiding excess inventory, so I'd evaluate both ML metrics like WAPE or RMSE and business metrics such as stockout rate, fill rate, and inventory holding cost. Since the product has no historical demand, I'd treat it as a cold-start problem and rely on features that are available before launch. I'd engineer features such as category, brand, price, supplier, launch region, planned promotions, seasonality, and product embeddings from text or images if available. I'd also create similarity features by identifying comparable products and using their historical demand as prior information. I'd train a feature-based regression model on existing products, where the target is demand and the inputs are these pre-launch features. Once the new product starts generating sales, I'd gradually incorporate SKU-specific features like lag demand and rolling averages, transitioning from the cold-start model to the standard forecasting model. For rollout, I'd first shadow the existing forecasting process or run an A/B test on a subset of new product launches. I'd compare forecast accuracy using WAPE or RMSE, but more importantly, I'd monitor downstream business KPIs like stockouts, fill rate, and inventory cost. If the new approach consistently improves those metrics, I'd gradually expand it to all new product launches. You will have to dive into more based on question from the interviewer - but this should give you a high level idea of topic to touch on. For more questions, checkout Decoding ML Interview - a prep resource with 100+ ML questions that you can actually finish before your interviews https://lnkd.in/gc76-4eP

If you're in manufacturing, you know that accurate demand forecasting is critical. It's the difference between smooth operations, happy customers, and a healthy bottom line – versus scrambling to meet unexpected demand, dealing with excess inventory and having liquidity issues, or losing out on potential sales and not meeting your Sales / EBITDA targets. But with constantly shifting customer preferences, disruptive market trends, and global events throwing curveballs, it's also one of the toughest nuts to crack. While often reliable in stable environments (especially in settings with lots of high-frequency transactions and no data sparsity), traditional stats-based forecasting methods aren't built for the complexity and volatility of today's market. They rely on historical data and often miss those subtle signals, indicating a major shift is on the horizon. Traditional stats-based approaches are also not that effective for businesses with high data sparsity (e.g., larger tickets, choppier transaction volume) That's where AI/ML-enabled forecasting comes in. Unlike foundational stats forecasting, it can include various structured and unstructured data, such as social media sentiment, competitor activity, and various economic indicators. One of the most significant advancements in recent years is the rise of powerful open-source AI/ML packages for forecasting. These tools, once the domain of large enterprises with extensive resources or turnkey solution providers (with hefty price tags), are now readily accessible to companies of all sizes, offering a significant opportunity to level the playing field and drive smarter decision-making. The power of AI and ML in demand forecasting is more than just theoretical. Companies across various industries are already reaping the benefits: • Marshalls: This UK manufacturer used AI to optimize inventory management during the pandemic. It made thousands of model-driven decisions daily and managed orders worth hundreds of thousands of pounds. • P&G: Their PredictIQ platform, powered by AI and ML, significantly reduced forecast errors, improving inventory management and cost savings. • Other Industries: Retailers, e-commerce companies, and even the energy sector are using AI to predict everything from consumer behavior to energy demand, with impressive results. If you're in manufacturing or distribution and haven't explored upgrading your demand forecasting (and S&OP) capabilities, I highly encourage you to invest. These capabilities are table stakes nowadays, and forecasting on random spreadsheets and basic methods (year-over-year performance, moving average, etc.) is not cutting it anymore.

Machine Learning-Powered Demand Sensing: Revolutionizing Real-Time Decision Making In the realm of demand forecasting, machine learning (ML) is reshaping the landscape by enabling real-time analysis for predicting short-term demand with exceptional precision. Unlike conventional methods that rely solely on historical data, ML-driven demand sensing incorporates a wide array of data sources, including sales figures, inventory levels, weather patterns, social media trends, and economic indicators, to swiftly identify fluctuations in demand. For instance, in the context of event management, demand sensing proves invaluable in anticipating attendance variations influenced by external factors such as weather conditions or concurrent events. Through sophisticated ML algorithms, subtle trends like a sudden spike in ticket purchases triggered by social media engagements can be detected, empowering organizers to promptly adjust their strategies related to inventory, staffing, or promotions. This innovative approach not only slashes forecast errors by as much as 50% but also streamlines resource distribution and mitigates risks associated with overbooking or inventory shortages. By translating raw data into actionable intelligence, demand sensing fosters agility and accuracy in navigating dynamic market conditions.

Ever wondered how warehouses can handle brand new products for which they have no historical order? Where do we place the article in the field to pick most efficiently? Will it be an A-article? Or will it be an absolute slow mover? We don’t know since there is no order data. But there is some information we can leverage. This new t-shirt from the new collection? Is it completely new? Well, yes. But we might have shipped a similar product in the past. And for this article we do have order data. First, we identify these “similar” existing articles. Similarity could be based on the price, the season, colors and design features. If we were an operation with a limited number of articles, say 100, it would be relatively easy to do this analysis by hand. I look at my new SKU and find the most similar item in the existing SKU base and take its historical order pattern as an indicator of how well the new item will ship. But how do we do this at scale? When we talk about apparel products, this problem becomes much harder very fast. We have to deal with potentially hundreds of thousands of SKUs and a lot of seasons or continued introduction of new articles. AI algorithms can analyze patterns across thousands of similar products and consider multiple factors simultaneously. To overcome this cold-start problem, we need to find the similarities.? How? We could for example embed product descriptions and other attributes into vectors and find the existing article that is closest in terms of a similarity measure such as cosine distance. Rather than comparing the new product to all existing products, we could also cluster. When early order data comes in for the new SKU, we can then adapt the produced forecast. And this is how we can handle SKUs in the warehouse that we have not seen yet. Follow me Dr. Jana Boerger and #datainlogistics for more content on data science in logistics and my path into the field. #datascience #logistics #datainlogistics #warehousing

Fastest way to lose $1M? Being OOS - You can fix your ads in a day - You can fix your pages in a week - Inventory mistakes last as long as your pipeline takes (for some categories that is 18 months) Compounding the actual lost sales in that time, your organic rank will also suffer, which will pull down your traffic and authority. * depending how long you are OOS this takes some time to build back How to avoid it? Forecasting First you need to make sure that you have separated what you want to be true from what is true. This is a first principle approach. Market: - What do my customers want? - How many of them? - What is the TAM (Total addressable market)? - What is the SOM (Serviceable obtainable market)? - What penetration does my brand have? - Break out by variant (color, size, flavor etc) Your brand: - What is my current share of voice? - How much of my traffic is paid? - How much of my paid traffic overlaps with market demand? Eg you promote the blue shirt so it's your top seller but the traffic to that page is 60%+ paid and the market (organic traffic and demand) is begging you for brown pants in a 38” waist. There is a delta here in what you wish were true vs what is actually true. 🚨 If you are a sub $20M a year brand this is going to become a huge cost for you. 🚨 For brands of all sizes this is a very large inventory risk. (this is how you get overstocked in the shirt and are OOS on the pants) If it diverges more than 30% separate paid demand from real demand before moving on. Tools: - If you have an Amazon account you should be running comparisons between P70 and P90 once a week so you can see the patterns in demand - You can use Google trends (its free) to run product search diagnostics AND phrasing diagnostics (eg there is a large difference in demand between “maternity pants” and pregnancy pants” you need to use both) From here you can start to build your reality based model. If you would like to see this in depth please grab the post in the comments. I'd love to hear your forecasting stories below 🙂 #ecommerce #amazon #reporting #analytics #forecasting #data