Transform Your Supply Chain Planning and Marketing Strategies with Google Cloud and SAP Integration

Cloud Marketplace

Real time demand forecast with ML and AI

July 29, 2020 | Divyansh Meena

Blog / Real time demand forecast with ML and AI

The unforeseen tension in the chain:

Imagine yourself holding a giant chain that spans across your network of suppliers and cuts through geographical boundaries. On one end it is controlled by you as the supply chain leader while on the other end, it is controlled by your end consumer. 

If your customers love your products, they pull the chain harder, pushing you forward. However, if they aren’t satisfied with the quality of your products, they might lose their grip on the chain, making you move backwards. It is true that in both cases you are trying to maintain the tension in your supply chain, so that it does not touch the ground. 

Given the new market conditions, probably you are not able to predict the tension in the chain for the first time in your experience. The fluctuation in demand patterns have seen peaks or dips that were never recorded in such a small time frame.

At Pluto7, we always have found solutions, and this time it is no different as demand forecasting has to be reimagined for retail businesses. We all agree the chain cannot lose tension and touch the ground. 

Answering questions while balancing demand during uncertain times:

There are questions all around us: “When are things getting normal?” ; “Who will win the vaccine race?” ; “ When will the tooth fairy arrive ?”  and all the nine yards. We are overwhelmed by questions from our family, team, and that flashy spammy email subject line that says “Do you even supply chain ?”. It is not the very nature or subject of these questions that take us for a ride, but the uncertain times. The demand for accurate forecasting in these uncertain times has skyrocketed. 

Pluto7 has time and again proven that forecast accuracy can be optimized if data patterns are closely watched rather than relying on human judgements and overrides. Equipped with domain knowledge and deep expertise working on Google Cloud Platform, Pluto7 team has been able to drive forecast accuracy by 95%+ — which was considered not possible in some cases. 

Currently, many organizations are moving from their initial response phase to the recovery phase, and they would slowly move through restoration and achieve the prevention stage (as shown by Gartner). 

Demand forecasting has to be re-imagined and here are five key elements that you can focus on to get started: 

  • Jump Start Innovation in demand forecasting. :

Companies need to ignite innovation and combine it with an agile supply chain strategy. It is important to include clear definitions of forecast accuracy and baseline values. Additionally, leaders should set expectations for existing and new products, including the time horizon for forecast and such key guidelines.

  • Enable real-time visibility into internal business and external data: 

To make your forecast more accurate, allow the team to centralize the data into a repository where they do not have to worry about storage costs or computation costs. Allowing a data scientist or a machine learning expert to enable key data points without technical constraints will help them identify new signals and trends that are not obvious.

  • Iterate with ML and AI technologies:

Raise the bar above the current norm for forecast accuracy and yet be realistic. Often we have seen customers get caught up with the ML model that gets used and test it. The ML model identification is the relatively easiest part and relative to the less complex part in the long run, but it will be harder to get the data in, processed, and get people buying to adopt ML over their own decision making.

  • Change management is more important than proving technology:

ML-based forecast accuracy is proven to be better over and over, across industries if you look in the right areas. Stock market, healthcare, space exploration has written thousands of models that are now packaged as ML models and algorithms in many white papers published by Google.

  • Blend business processes with newer ML/AI technologies:

Augmenting and not replacing humans should be the first approach so that the humans can help achieve extraordinary forecast accuracy not seen as possible before, and run 100s and 1000s of scenarios with ML models to simulate the business forecast in the “New Normal”.

blog images

Build stronger and more stable supply chains

It is suggested that you take a deeper look at the discussion by reading this free whitepaper drafted by our AI experts: Leverage real-time demand forecasting to build an agile Supply Chain and optimize business continuity.

Companies of all sizes need to have an ability to plan for various scenarios while bringing agility into their supply chain. It sounds easy, however, experimenting with new technology can always be a challenge — and that is where we can help!

We would love to hear from you, please send all your queries to