Building predictive analytics in an unpredictable age
Predictive analytics are a critical business tool for supply chains but producing effective forecasts is especially tough in this unprecedented environment says panel of experts
With wild demand fluctuations and consistent disruptions to both production and transportation, supply chain managers are in dire need of accurate modelling and predictions. The problem is that the black swan event of COVID-19 has swept historical precedent away and thrown out data that could have been relied on for a typical year. This is making predictive analytics a tough proposition this year for all players working in the logistics space.
To find out where the issues lie and how companies are adjusting, Reuters Events: Supply Chain convened a panel of experts in a recent webinar.
A wildly unpredictable era
Critically, “Predictive allows us to get ahead of the demand,” said Cindy Headen, Executive Vice President Supply Chain for Domino's, which “allows you to prepare in advance all across your supply chain: Your labour force, your make, move, and sell - all of that predictive really enables, and allows you to be a lot more proactive.”
It's hard to be predictive, [as] that history is no longer a predictor of the future
This is a valuable asset for any company to have at any point but especially right now, when there are big knock on effects from upstream disruption and major fluctuations constantly going on in demand patterns.
However, it relies on prior data being a reliable source to call back on and right now those unusual conditions means “It's hard to be predictive, [as] that history is no longer a predictor of the future,” pointed out Headen.
For Erik LaValle, Digital Supply Chain & Customer Experience Technology Portfolio Leader of T-Mobile the changes are especially profound, as “You can think of it as the butterfly effect. Something that happens upstream can have a significant effect downstream. A small change in a plan, or a forecast may have big impacts on suppliers, on physical transportation, etc., or a customer event can have that as well. So, it's the complexity of all of the elements that can come in and influence what happens in your supply chain operation.”
This effect would normally be caused by a single supplier struggling to fulfil or a single market being affected by an event, but right now the complexities are everywhere and the changes in dynamics massive, stretching out across continents. “In this case, it is hard to regionally business continuity plan a pandemic,” says Headen, “because if the nation goes down or your supply base goes down, you now have a national crisis, not a regional localized one.”
We had multiple orders in the 300,000- to 600,000-unit range for things like school systems, Los Angeles, New York, etc. As you can imagine, for a supply chain that looked like … a snake swallowing a bowling ball. That's just not a normal demand pattern for us and was completely unpredictable
Alongside supply issues, demand patterns are deviating wildly from their norms. Greg Orr, President CFI, EVP US Truckload, TFI International, found that “35% to 40% of our business that we actually managed on a continual basis in Q1 was not there in Q2, so we had to figure out how to adjust a quarter accordingly to make sure that the providers that are still shipping."
“When COVID hit my company, for example, the demand patterns completely shifted, not only where people would ask for product, but the levels, volumes and sources of that demand,” explained LaValle. “We had multiple orders in the 300,000- to 600,000-unit range for things like school systems, Los Angeles, New York, etc. As you can imagine, for a supply chain that looked like … a snake swallowing a bowling ball. That's just not a normal demand pattern for us and was completely unpredictable.”
Working and adapting in a tough environment
All of this means that “Getting to predictive has been very challenging,” says Headen.
This has called for new strategies to cope with the situation. For Orr, the key is to become “More adaptable, more flexible, and be able to adjust accordingly. I think that's been the biggest takeaway from a transportation providers view is that ultimately, we have to be able to adjust to what our customers needs are.”
So, whether it was a different place to get that product from, [or] a different product that met the need for that demand, and some creative ways of doing logistics and operations to move that around, we were able to substantially meet the demands that we had that were completely unanticipated
Part of that has been “Evaluating not only the existing supply that was available but also alternate supplies. So, whether it was a different place to get that product from, [or] a different product that met the need for that demand, and some creative ways of doing logistics and operations to move that around, we were able to substantially meet the demands that we had that were completely unanticipated.” This relied on “visibility of knowing where we had both existing supply and potential supply, [and then] the collaboration that we had with those suppliers.”
For Headen, the adaptability comes in the forecasting as well, shortening down the timeframe for data inputs and adjusting the weightings to reflect the current circumstances. Domino’s are now looking at their data “as a subset of the last four weeks” rather than “what has happened the past six months.” They also “probably err on the over versus the under in proactive planning at this stage of the game. We would rather be over prepared, whether it be from a supply demand perspective, than err on the under.”
Alongside adjusting forecasting methodologies and focusing in visibility in order to undergird adaptability, LaValle emphasises the importance of putting in the fundamentals for understanding data. Visualisation is key, particularly, “Real time dashboards,” that can help with “immediate visibility of complex things that are happening kind of behind the scenes, especially “as you begin to combine data in more sophisticated ways.”
The end result
Predictive analytics is a fusion of different elements, requiring a decent degree of visibility across supply chains, digital capacity, real time data feeds (especially in the current environment) and an ability to manipulate and break down data to reflect the current environment. These feed into a wider picture of supply chain effectiveness: “When you're able to see a supply chain in real time, and you're able to problem solve problems real time, then you should really have last mile,” said Kendra Phillips, CTO, Supply Chain Solutions for Ryder System, Inc.
I really see digitizing the supply chain as resulting … [in] shorter movement of goods, fewer miles driven, less returns, and this will just continue to create efficiencies
Building these capacities means “You should be able to control better the freight movements, ensure they're up in an optimal position, and that you're getting into the right point of sale, [so] you should be able to optimize the use of trailer space…. I really see digitizing the supply chain as resulting … [in] shorter movement of goods, fewer miles driven, less returns, and this will just continue to create efficiencies.”
Orr gives the example that, “On average, Drivers sit anywhere from 15 to 18 hours a week in dwell time, meaning that they're either waiting for their next load or waiting to be offloaded. If you just think about that amount of time across, call it a 2,000 truck fleet, that's over 30,000 hours a week. That is wasted, that's non-productive for the drivers and/or the equipment, but secondarily, you know, spinning off some type of carbon emissions to the environment.”
This means “From our standpoint, the more that we can have that predictive environment … the better off we're all going to be … a little bit more green than what we are today.”
To hear Scott Sureddin, CEO of DHL Supply Chain North America, talk about the topics and issues covered in this feature, along with many more expert speakers, join our online event - Supply Chain USA Virtual this October 7-8th.