Top

Would you like to:

TALK TO SALES? JOIN OUR TEAM?

Helping a leading fresh food supplier successfully predict demand during the COVID-19 pandemic

Key highlights
  • Provided 24 combination-specific future projections at category, channel and region level
  • Developed a dynamic and self-service scenario modelling simulator for what-if insights
  • Delivered demand forecasting predictions with up to 95% accuracy
Download a copy

Business challenge

Effective demand planning is one of the foundational capabilities of revenue growth management. But with COVID-19, unprecedented volatility made gaining insight into demand extremely difficult. For many organisations, it’s been next to impossible to make the right products available through the right channels at the right time.

Our client is a fresh food supplier that strives to promote a plant-based diet in the United States. Its long-term goal is to help improve the health of the planet – and the people that populate it.

One of its major product lines has provided American consumers with healthy, fresh, ready-to-eat food bowls for nearly fifty years. This wealth of experience, supported by a team dedicated to forecasting demand, meant the company was well placed to negotiate its market. But with the pandemic causing wide-scale disruption, it knew it would need a deeper level of insight to weather the storm.

Operating across four categories, four channels and three regions, the company approached The Smart Cube looking to gain predictive insights into 24 different combinations for effective commercial, sales and operations planning.

The client required a solution that could easily be shared across finance, commercial and supply chain teams, and provide insight to senior managers, development directors and procurement professionals alike. What’s more, it needed the agility to plan for both the short and medium term, which meant forecasting demand for the next three months, as well as the 15 beyond that.

Ultimately, it wanted to gain as much visibility as possible into the variables contributing to such an unpredictable environment. See how we helped a major fresh food supplier accurately predict demand in the short and medium term during COVID-19. 

The Smart Cube solution

Deploying the demand planning module of our Revenue Growth Management solution, our first step was to augment the client’s shipment and events data for the last three years with external data sources related to macroeconomic trends and the current pandemic.

This included:

  • Point of Sale (POS) data from Nielsen
  • Channel and shopper data from Numerator
  • Macroeconomic data from the US Bureau of Labor Statistics, Bureau of Economic Analysis and US Census Bureau u COVID-19 data from IHME (Institute for Health Metrics and Evaluation), Centers for Disease Control and Prevention (CDC), Google Mobility Report and the University of Oxford’s Stringency Index
  • Consumer Confidence Index data from the Conference Board and CSI (Consumer Sentiment Index) by the University of Michigan.

Our next task was to harmonise, clean, standardise and process this data, and understand the best way to merge it to provide the most accurate insights.

We then set about a process of feature engineering, extracting information that wasn’t readily available in that dataset to gain new insights into variables. This involved creating new fields like average selling price from internal data and COVID rates, creating flags to signify unexpected events, introducing features to capture lag effects associated with events, and performing aggregations to obtain a consistent level of data granularity.

The nature and scope of this project meant it was more than just a simple forecasting exercise. As well as predicting volume over the next 18 months, the client wanted to gain insight into the different variables that could impact anomalies. For instance, if there’s a big dip in sales, they wanted to know specifically what had caused it. This meant we had to take a parametric approach to forecasting – allowing the client to analyse numerous variables at once.

Delivering this insight required ameliorating standard forecasting techniques by combining time-series forecasting and regression-based modelling.

However, when the project started, understanding how COVID-19 impacted each variable and how that affected the business wasn’t simple. At that point, we only had 12-16 weeks of COVID data to work with – a drop in the ocean in the grand scheme of the pandemic.

For this reason, we created three different scenarios based on the patterns we expected the pandemic to follow: one of which would see infection numbers decline in October, another would see them decline in January, and a third predicted recovery would begin in April. These scenarios were updated and refined as we learnt more about the evolution of trends in the pandemic.

Results

The solution created for our client is now used by several key stakeholders in different departments across the organisation for integrated planning. It provides access to 24 different combination-specific future projections, with data refreshed every month to ensure the most accurate and timely insights to aid the Sales and Operations Planning (S&OP) process.

In addition, each of our client’s teams now has access to a self-service simulator, allowing users to adjust a range of variables, simulate conditions and prepare for any number of possible circumstances. The simulator provides insight into dollar and volume sales performance across any combination of categories, channels and regions.

The tool also allows the client to track how close actual results are to forecasts, further instilling confidence in its data-driven decision making. And the results have fed straight into the client’s budgets for the next year, with finance requesting forecasts until June 2022.

What’s more, the solution we developed is infinitely scalable, should the client want to use it for other categories and geographies in the future.

Value generated

With the data and insights provided by The Smart Cube, our client’s supply chain and commercial teams are now able to construct strategies based on demand forecasts that achieve up to 95% accuracy.

And, with regular reporting on previous forecasts and how they aligned with real-world events, these teams have the assurance they need to tackle an unpredictable market with confidence. 

“The Smart Cube enabled us to gain insights into a market that is as difficult to predict as you can imagine,” said one executive. “We now have a clear idea of demand, what impacts it, and how to strategise accordingly”.

Learn more about The Smart Cube’s Revenue Growth Management solution to see how our comprehensive insights can help you maximise revenues and drive profitable growth. 

Privacy Preference Center

Necessary

Information collected using our website: When you contact us or send us an email, we may collect such personal information that you freely supply to us. We also collect any personal information that you post publicly on the website. Our server logs automatically collect information, such as your IP address, your browser type and language, the pages on the website visited, and the date and time of your visit, which helps us gather statistics on the website and to track users’ movements around the website in order to customise and improve the website and your experience of the website. We may use standard internet tools, such as web beacons and cookies, which collect information that track your use of the website and enable us to customise our services and advertisements.

Advertising

Analytics

This cookie is placed when you click ‘Accept’ on the cookie information warning bar that is displayed at the foot of our website and tells us you have given your consent to the use of cookies on our site and stops this message from appearing. A compliance cookie is also placed if you change your cookie preferences using the control panel on our website.

Other