How Do You Improve Demand Forecasting Accuracy by 20% with External Datasets Using SAP IBP and Google AI Cloud?

Blog / How Do You Improve Demand Forecasting Accuracy by 20% with External Datasets Using SAP IBP and Google AI Cloud?

Traditionally, demand forecasting has been rooted in the analysis of historical sales data. While this approach offers a baseline, it’s inherently backward-looking and often fails to account for the dynamic factors that influence future demand. These traditional models struggle to adapt to rapid market changes, seasonal variances, and external shocks, leading to inaccuracies that can ripple through the supply chain, causing overstock, stockouts, and lost sales. 

The core limitation lies in the assumption that past patterns are reliable predictors of future demand, an assumption increasingly challenged by the volatile nature of today’s markets.

Let’s take the example of a demand planner using SAP IBP as their primary planning tool. They want to forecast demand for a newly launched line of heart-shaped pendants and their top-seller engagement rings in the Los Angeles market for Valentine’s Day, which is also their biggest sales season. 

⚠️ But here is the problem – they need to predict which jewelry styles will become more popular in Los Angeles by analyzing online searches and social media mentions before placing final orders for materials. 

• Social media trends → External datasets (How can I integrate external data sets into SAP IBP?)
• Historical sales data → Salesforce (How do I import last year’s sales data from Salesforce into SAP IBP?) 
• Supplier availability → Supplier Management System (How quickly can I update SAP IBP with real-time information on material availability from suppliers?)
• Current in-store favorites → Retail POS (What’s required to link our POS system with SAP IBP to reflect real-time sales data for our LA stores?)
• Customer design feedback → Customer Feedback System (Can I set up a direct feed of customer feedback into SAP IBP to adjust forecasts based on real-time sentiment?)

Example Scenario: A Retail Company Enhances SAP IBP With External Datasets 

ACME Retail, a jewelry retailer specializing in diamond rings, operates across the US market through retail chains and e-commerce platforms. Valentine’s Day is their most significant sales event, they aim to leverage Google Search trends data from specific US regions to enhance their demand forecasting accuracy. Here’s how they achieve this:

How to add Google Search Trends Data into SAP IBP? 

ACME Retail uses Pluto7’s Demand ML, a machine-learning solution hosted on Google Cloud. Demand ML offers a flexible architecture that allows ACME to plug external datasets, such as Google Search trends and weather data, directly into SAP IBP. 

A real example of a Demand Sensing Dashboard implemented by Pluto7. Learn more here. 

Step-by-Step Process:

• Integration of Google Search Trends & Adtech Data: ACME starts by integrating regional Google Search trends & Adtech data for keywords related to diamond rings and Valentine’s Day. This data helps ACME understand consumer interest and potential demand spikes in various US regions.
• Incorporating Weather Data: Next, they add weather data to the model, considering how weather patterns can affect consumer behavior and shopping trends, especially in their e-commerce channels.
• Demand Forecasting with Demand ML: Using Demand ML, ACME processes this integrated data to generate demand forecasts. The machine learning algorithms analyze patterns and correlations between search trends, weather conditions, and historical sales data to predict demand more accurately.

How Effective Are Demand Sensing Algorithms?

A real example of a Demand Sensing dashboard that plugs in weather data along with 250 other external datasets.  Learn more here. 

Our internal data shows that over 90% of companies using Demand ML have improved their forecast accuracy by 30-40% within 4-6 weeks. To illustrate the improvement in demand forecasting accuracy, let’s consider hypothetical numerical data:

Is It Possible to Make Adjustments to the Data Using Spreadsheets?

With Pluto7’s Demand ML, and similarly across Pluto7’s decision intelligence platforms, users are granted complete control over data models, UI, and dashboards. For ACME, this means: 

• ACME’s planners can slice and dice the data by region, product, and time period.
• They can export the forecast data into spreadsheets, where they make final adjustments based on their market knowledge and recent marketing campaigns.
• As new data comes in, ACME continually updates its forecasts, ensuring they remain as accurate as possible up to and during the Valentine’s Day sales period.

This capability is crucial for demand planners who prefer to perform detailed analyses or adjustments in a familiar spreadsheet environment before finalizing their demand forecasts.

Do You Want to Bring the Power of External Datasets into Your ERP?

Throughout this blog, we have explored multiple scenarios where integrating external data proves invaluable. When we work with companies, they highlight three key benefits that significantly improve their data handling:

• Automated Data Integration: By automatically bringing in external datasets and continuously updating the models with new data, decision-making becomes more informed and responsive.
• Connecting Systems for Complex Analysis: Enterprises appreciate being able to link everything from ERP to CRM with external data, allowing them to dig deep into complex analyses without getting tangled in multiple systems.
• Streamlining with Gen AI: The shift towards using Gen AI cuts down on the need for writing lengthy SQL queries. This means getting to the insights you need quicker and with less effort, using natural language to query your data.

The ability to harness all these benefits, packaged efficiently within a 4-week timeframe, is why companies like AB InBev, Levi’s, Ulta Beauty, and others have chosen us for their data transformation projects. Interested in seeing how our decision intelligence platform can enhance your ERP data? Reach out for a personalized workshop.