by Peter Sayer

Lennox enlists BI to anticipate COVID-19 supply chain disruptions

Jul 13, 20206 mins
Business IntelligenceData VisualizationIT Leadership

With the pandemic significantly upending global supply chains, the HVAC manufacturer quickly rolled out a BI dashboard to predict, respond to and get ahead of potential problems.

abstract digital network mapping concept / virtual globe of connections
Credit: MetamorWorks / Getty Images

The COVID-19 pandemic has profoundly disrupted supply chains across the globe. This is especially true in the manufacturing industry, where supply chain managers must track and analyze the pandemic’s impact in every region that effects their business: Is it safe to bring staff into factories? Are suppliers still shipping — or even still in business? Are customers still buying? Can distributors still serve them?

Richardson, Texas-based Lennox International is one such organization that depends on a complex global supply chain. The heating, ventilation and air conditioning equipment supplier has manufacturing facilities in Arkansas, Georgia, Iowa, Mississippi and South Carolina, and its network includes regional and local distribution centers and stores. But it also sells products through dealers and contractors, meaning there are plenty of ways the pandemic can upend Lennox’s ability to deliver its products to its customers.

As the coronavirus swept the country, Lennox figured there had to be a better way than having staff scour news sites to find out how the disease was spreading in the communities around the company’s factories and distribution centers, or how suppliers and key customers were faring.

“Like everyone, we had our fair share of challenges that was impacting our key KPIs,” says Satish Seshayya, IT business applications leader at Lennox. So, in early April, as COVID-19 cases hit their first peak in the U.S., Lennox IT decided to map out the company’s supply chain network and look for potential disruptions that could affect order fulfillment or customer service, presenting them to management on a dashboard.

Lennox already relies on an array of reporting tools. It runs its business on SAP – it’s a reference customer for SAP’s global demand planning tool – and uses other applications to manage supply chain operations too. But the team quickly realized none of Lennox’s software suppliers could come up with a solution that would highlight and predict the kinds of supply chain disruptions they were concerned about without a major project undertaking or access to internal data for other companies.

Rapid development

Negotiating that access might have been possible — but with the pandemic unfolding, the dashboard was something that needed to be developed rapidly: Its value would be lost if it turned into a six-month project.

Satish Seshayya

“We approached it as an internal development effort using our internal data scientists and resources. We had the tools and resources to develop a solution using our data,” says Seshayya, who pitched the idea to the leaders of the affected business segments and the supply chain staff involved, to identify how best to present the data.

“As with any product development, you always poll the customer to get feedback,” he says. “We engaged all the supply chain folks that plan and handle deployment of materials both upstream and downstream.”

Developing a prototype to present to the business side, then iterating on it to make improvements based on their feedback, took about three weeks.

“We took a DevOps approach with a few sprints to receive continuous feedback from the business,” Seshayya says.

Four people were closely involved in the project: Seshayya and a colleague on the functional side, and a data scientist and a developer on the technical side.

The biggest difficulty he and his colleagues had to overcome was getting hold of the data to accurately predict disruptions. There were plenty of external sources providing state-by-state or country-by-country structured data sets of COVID-19 case reports and projections, including the Centers for Disease Control (CDC) and the World Health Organization.

“Internal data was challenging — particularly supply source of materials,” he says. “We got creative in making some assumptions with our supplier data that helped solve the issue.”

Early warning system

Combining all that in Microsoft Power BI with some machine learning tools, Seshayya and his team were able to offer their colleagues an early warning system for potential supply chain problems in areas where COVID-19 cases were increasing or expected to increase in the next couple of weeks.

“As you feed information into it, it starts learning from it, and it starts making predictions in terms of this particular location is going to be most impacted, or that location,” he says.

Lennox has already used machine learning to measure and forecast failure rates for some of the products it sells, or to help it better manage warranty costs, for example.

The early warning system provides various views by region for the upstream (supplier) and downstream (customer) supply chains, grouping the factories or fulfillment centers into color-coded categories for low, moderate or high risk. Supply chain managers can drill down to see which materials or deliveries will be affected and can see at a glance whether they will need to seek alternative sources for materials or fulfill orders from a more distant facility.

“This model has generated a whole new set of thoughts and ideas on managing supply chain disruption before it enters the red zone,” he says. 

It has also helped the business identify potential pitfalls through simulation. “What-ifs have helped tremendously and assisted in diverting the fulfillment sources, including supplier bankruptcy risks,” he says.

If Seshayya has one regret, it’s that they didn’t come up with the idea sooner — before the problem became a pandemic.

Lennox has had good usage out of its new supply chain dashboard all the same, and while we can hope that the threat of COVID-19 will soon recede, as it has in other countries, the dashboard won’t become redundant.

“We plan to repurpose the same model and solution to other natural disasters to predict disruptions,” Seshayya says.

More on supply chain management: