Crossing the Great Forecasting Divide Part 2 – Consumption Sensing is New Forecasting Frontier

    

In my last post, I focused on how retailers and consumer products companies can triage through the worst of the COVID pandemic as consumers were adjusting to a stay-at-home lifestyle. I discussed using a hybrid model of humans and AI/ML-based approaches, delivering science-as-a-service, to bridge the gap and react quickly to the dramatic swings in demand.

Three weeks have passed since then and we face a new paradigm – the re-opening of society and our economy. It is still early days, no doubt, but what is certain is that traditional forecasting solutions that rely on historical data are not going to be of much use for the foreseeable future. The new “table stakes” in forecasting will be using an outside-in view, focusing on what consumers will want and need in the next days, weeks or months. In other words, sensing future consumption, whatever it may be, rather than repeat history.

So, what is Consumption Sensing? It is an AI/ML-based forecast that combines new sets of data such as exogenous variables, syndicated data, weather, inventory, local events, promotion tactics, daily demand across all channels, and yes, historical demand transactions – it does have a role in the new paradigm. 

Consumption Sensing understands demand patterns in the short, medium, and long-term horizon by intelligently modeling the confluence of available data sets to predict the future.

No longer does a single, independent variable drive the forecast, but rather a collection of demand drivers come together whereby the AI/ML models can ascertain the degrees of importance of those demand drivers, individually and as a whole, to create the best possible forecast over days, weeks, months and years.

With a lens into the future, Consumption Sensing can be used for practical use cases, such as: 

Short-Term: Creating a responsive demand signal that can drive daily and weekly replenishment, production and inventory optimization decisions.

Medium-Term:  Providing a critical input for S&OE and S&OP to balance demand and supply across the decision time frame and granularity of execution as defined by time, product, market and analytical hierarchies.

Long-Term: Delivering Commercial ROI by understanding the intersection of consumer consumption behaviors, trade spend and shelf execution to optimize spending investment actions.

The baseline of demand has shifted, and just like two tectonic plates over a fault line can create a massive earthquake because they don’t align, the same analogy can be used when we refer to the baseline of pre-COVID to the baseline of post-COVID. This seismic shift in forecasting requires us to embrace newer technology to fully leverage the array of data available to improve decision-making – from the supply chain to revenue growth management to account planning and S&OP/IBP.