Philip Morris International’s Pivot To Smoke-Free Products Complicates Their Supply Chain

The tobacco industry is changing. Traditionally, the industry has been stable. The prevalence of smoking is going down, but the world’s population is increasing; the net result is that the forecast for the number of smokers is essentially flat in 2025 as compared to 2010. Because of this stability, at Philip Morris International (PMI

PM
), one of the world’s leading international tobacco companies, budgeting out three to five years used to be something they could comfortably do on spreadsheets.

Dr. Alexandros Skandalakis – the Director Global Manufacturing Capacity, Strategic Assets and Capital Expenditures at Philip Morris Products S.A. – explained that when conducting manufacturing volume forecasts for budget purposes, the company would look at demand, production capacity, product specifications, seasonality, asset location, costs/duties and new product introduction amongst other criteria to satisfy regional volume demands. This was done at a stock keeping unit level and for the entire manufacturing supply chain. At the end of 2019 that supply chain covered 38 PMI owned factories, 28 third party manufacturers, and more than 180 markets. Once the analysis was done for Year One set up, Year Two was pretty much the same. It was predictable.

But in 2016, Philip Morris International decided to change the course of its history by leading  a  transformation in the tobacco industry to create a smoke-free future based on products, which while not risk-free—are a better choice for adult smokers than continuing to smoke. In July of 2020, the US Federal Drug Administration (FDA) authorized the marketing of a version of IQOS, a heated tobacco product that Altria

MO
is selling under a license agreement with PMI in the US. The FDA issued an exposure modification order  that allows the claim to be made that “scientific studies show that switching completely from conventional cigarettes to IQOS significantly reduces your body’s exposure to harmful or potentially harmful chemicals.”

The introduction of smoke-free products made the use of spreadsheet tools far less efficient in the capacity and sourcing planning as the new product categories had rapid growth. “We needed to simulate multiple scenarios of significantly increased capacity and sourcing complexity, at speed and over a long planning horizon. We realized very quickly that our legacy systems and traditional approaches were becoming obsolete. We needed to re-invent the way we work fast and at scale to become fit for the future”, said Dr. Skandalakis.

What PMI needed, considering the long planning horizons, was a digital and analytics network design and supply optimization tool. Network design and optimization tools help companies understand how to best leverage enterprise assets across their supply chain networks and where their production volumes should be optimally located. The solution considers projected demand and service level goals. Those goals are balanced against a variety of costs and parameters from the manufacturing supply chain. The objective is to meet the service level goal with a network design that optimizes the costs across the manufacturing supply chain.

PMI looked at four different solutions in late 2018.  They ended up selecting a solution from River Logic. Dr. Skandalakis said the decision was based on many considerations and not solely on cost. “We appreciated River Logic’s capabilities.” The tool was able to create a model going out multiple years. If changes were made to any of the model parameters in Year One, those changes would ripple down through the subsequent years seamlessly, resulting into a cost optimal plan over a 10 years horizon. There was no need to change parameters or re-run models in any subsequent years. This made running multiple scenarios and keeping track of assumptions along with the proposed implementation plan(s) far easier.

PMI was not an easy prospect. As they heard promises about how easy it was to create models using various digital & analytics tools, they said “prove it.” They gave River Logic cost, lead time, and other data on five factories, and requested they built a validated network model as soon as possible. River Logic came back with a working and validated model in less than one week.

Following the River Logic selection in February 2019, work started on building the PMI’s Global Manufacturing Digital Twin model at scale. Real manufacturing and supply chain parameters, product volumes and specifications, material and product flows, network constrains and financials were captured amongst others criteria to ensure that model behavior would replicate reality correctly. Variables around decisions relating to market and factory sourcing, options to release or re-deploy assets, costs and others were added.

 “We underestimated the effort in the beginning as data collection and standardization proved to be the hardest part of the implementation” noted Evgenii Shekhter – Manager Footprint, who led the implementation on PMI side. The project rollout involved several steps, which were broken down in 3 categories. The first entailed the model development and data input, the second entailed the user interface design and implementation along with user onboarding and training, and the third was developing and redeploying new ways of working across the organization. By December 2019 the system and model were fully live.

They took data from multiple sources across 70 distinct input tables that included fixed and variable manufacturing and distribution costs, the location of the manufacturing sites, machine and product specifications, operating efficiency (OE), duties and taxes, capacity by asset and format and other important parameters defining the global manufacturing and product sourcing network. They put significantly more parametrized data variables in the River Logic model than what they were using with the Excel models. The model enabled handling massive variability in sourcing decisions, asset allocation decisions, footprint extension and conversion decisions, to name a few, across a 10Y planning horizon. On top, model design enabled simple setup of business constraints such as regulatory, operational and others. This allowed them to move from a 2 year to a 10 year planning horizon while accommodating the increased complexity and at scale. Analysis and scenario formulation lead time went from weeks to days.

Scenarios could be run on each of the mentioned parameters. What if instead of an average OE of e.g. 70% at this factory, it went to e.g. 75%, what would that mean to the business? What would happen if volumes significantly increase in a given market or region? What are the actions necessary to ensure uninterrupted supply? Dr. Skandalakis mentioned that “with our Global Manufacturing Digital Twin, we could stress test any design parameter influencing our Global Network and get results even in just a few minutes”.

The PMI Global Manufacturing Digital Twin is the reference model for how their manufacturing network runs on a long term horizon. The model is updated quarterly with tight permissions around who can do the updates and what types of data can be changed.

When I asked Dr. Skandalakis about the ROI, he said “we are talking payback in couple of hours.  The cost savings opportunities identified run into the hundreds of millions of dollars over the period examined.”  

Not every scenario is executed.  “The company must decide on the options and examine the opportunity costs.”

The project has been so successful, the company is now looking to model other parts of their end-to-end supply chain. They are expanding their digital twin approach to in-market distribution optimization and raw material purchases.

I asked Dr. Skandalakis for a conclusive statement with regards to the transformational change they achieved. He mentioned that “change is an essential part of a healthy business, we needed to learn to unlearn and learn again fast. The journey was not easy but the price of not changing would have been huge.”