By Paul R Salmon FCILT FSCM FCMI
Published for the Supply Chain Council and CILT
Introduction
In the modern world of analytics, digital twins, and predictive modelling, it’s easy to forget that some of the most profound insights in operational research (OR) emerged not from sophisticated software, but from careful thinking under pressure during wartime. One such story – the tale of bullet-riddled aircraft returning from missions in World War II – offers lessons that continue to resonate far beyond the battlefield.
This historical case not only laid the groundwork for how we think about survivorship bias but also holds powerful relevance for today’s supply chain professionals. Whether we’re forecasting inventory needs, mitigating risks, or optimising logistics, the central message is clear: what you don’t see can hurt you.
The Problem: Planes Were Being Shot Down
During the height of World War II, the Allied forces were suffering severe losses in their air campaigns. Military planners wanted to increase the survivability of aircraft flying bombing missions over enemy territory. A logical approach emerged: study the damage patterns on the planes that made it back from missions, identify where they were hit the most, and reinforce those areas.
Data began to be collected. Statisticians and engineers recorded bullet holes and shrapnel damage on returning aircraft, noting that certain areas – notably the fuselage, wings, and tail sections – consistently showed signs of impact. These were, naturally, assumed to be the weak points.
The first solution proposed was simple: add armour to the areas with the most bullet holes. More protection should equal fewer losses, right?
The Insight: Survivor Bias and Abraham Wald
Enter Abraham Wald, a Hungarian-born mathematician working with the Statistical Research Group (SRG), a team of civilian researchers supporting the US military. Wald examined the same data but reached an entirely different – and deeply counterintuitive – conclusion.
Wald realised that the analysis only considered planes that had survived. The bullet patterns were not telling the full story — they were only showing where a plane could take damage and still return to base.
His pivotal insight: The parts of the plane with little to no damage were likely the most vulnerable. Damage to those areas (such as engines or the cockpit) likely meant the plane was lost – and thus never studied.
In modern terms, this was a case of survivorship bias – the logical fallacy of drawing conclusions from an incomplete dataset, biased by the very process of selection.
Wald’s recommendation: don’t reinforce where the bullet holes are. Reinforce the areas that are missing from the damage maps. These “silent zones” were not untouched because they were safe – they were fatal when hit.
A Paradigm Shift in Operational Research
Wald’s logic flipped the thinking of the day. Rather than trusting surface-level data, he applied deeper systems thinking and statistical reasoning to understand the unseen. This moment became a foundational case study in Operational Research.
The key lessons?
Don’t only analyse what’s visible Understand why certain data might be missing Think about the system as a whole, not just the surviving outputs
Wald’s approach helped save lives by improving aircraft design and armour placement – all without firing a single bullet or running a single simulation.
From Combat to Commerce: Relevance for Supply Chains
So, what does this have to do with modern supply chains?
Plenty.
Supply chains are complex systems where decisions are made under uncertainty, and outcomes are shaped by invisible factors. Just like in wartime OR, the data we see is often incomplete. We analyse what’s available — the shipments that arrive, the stockouts we notice, the suppliers that perform well. But too often, we ignore the failures that never appear in the data.
Let’s look at how this plays out in several key supply chain scenarios.
1. Inventory Performance and “Quiet” Products
Many organisations focus improvement efforts on the products that generate complaints, stockouts, or excess cost. These are the ones that show up in dashboards and attract attention. But what about the parts that seem to perform perfectly – those that never cause problems?
In some cases, that’s because they’re working well. But in others, it’s because they’re not being used, are no longer relevant, or are so underreported that they hide hidden inefficiencies.
Just as the undamaged areas of returning aircraft weren’t safe zones, “perfect” inventory items might be untouched for all the wrong reasons – obsolete, under-utilised, or unmonitored. True optimisation requires questioning the silence.
2. Supplier Risk and Hidden Dependencies
When assessing supplier performance, organisations often rely on vendors with a proven track record — the ones that delivered well in the past. But what about the suppliers that were never chosen?
In some cases, they didn’t meet requirements. In others, they may have been more resilient, more local, or more sustainable – but overlooked due to procurement inertia. If your current supply base is shaped only by historical survivors, you may be reinforcing survivorship bias and locking in fragility.
The wartime insight would be to ask: what supply chains failed silently before they were ever built? What options didn’t survive long enough to be measured?
3. Disruption Reporting: The Dangers of “Success Stories”
Supply chain disruption reporting often centres around recovery: “We bounced back from a cyberattack in 36 hours,” or “We rerouted around a blockage within the week.” These are valuable insights – like the planes that came back.
But what about the disruptions that didn’t result in a bounce back – or those that were never reported because the business line failed altogether?
Ignoring these “non-survivors” – the failed pilots, failed expansions, or lost customers – skews our resilience picture. As with Wald’s aircraft, the gaps matter more than the hits.
4. Forecasting Demand: Beware of the “Seen” History
Many demand forecasts are based on historical usage. But what if usage is constrained by availability, awareness, or affordability?
Forecasting based only on past demand is like reinforcing only where the bullets hit. It assumes past consumption equals true need, ignoring hidden demand that never manifested due to supply failure or poor visibility. Organisations risk under-forecasting precisely where they are most vulnerable.
How to Avoid the Survivorship Trap in Supply Chains
So, how do modern supply chain leaders avoid falling into the same trap that almost misled wartime analysts?
A. Build Systems Thinking into Your Analytics
Don’t just look at the outcome – look at the system that produced it. Understand the inputs, exclusions, and assumptions behind your data. Challenge what’s missing.
B. Hunt for Negative Space
When conducting reviews or root cause analysis, deliberately look for absences: suppliers you didn’t choose, customers you lost, products that didn’t sell. These are your “missing aircraft.”
C. Model Counterfactuals
Good OR isn’t just about what happened, but what could have happened. Scenario modelling and stress-testing can surface unseen risks that traditional dashboards miss.
D. Balance Quantitative and Qualitative Data
The bullet hole story reminds us that raw data can be misleading. Human insight – like Wald’s reasoning – is essential. Include narratives, interviews, and frontline perspectives in your analysis.
E. Ask Better Questions
Instead of “What failed?”, ask “What didn’t make it far enough to fail?” or “What data are we missing because of our process?”
Conclusion: Why the Planes Still Matter
Operational Research was born in war, but its lessons are timeless. Abraham Wald’s insight didn’t just save aircraft — it saved the field of analysis from itself. By understanding what the data didn’t show, he revealed a universal truth: the greatest risks often lie in the gaps, not the graphs.
In the 21st-century supply chain, with its torrents of data, AI models, and digital dashboards, it’s tempting to assume we see everything. But Wald reminds us: the absence of evidence is not evidence of absence.
True resilience, agility, and optimisation demand that we look harder, think deeper, and never stop asking:
“What are we not seeing?”
About the Author
Paul R Salmon FCILT FSCM FCMI is a senior logistics and data strategy professional with extensive experience across defence, government, and industry. He is Chair of the CILT Defence Forum and a recognised advocate for data-driven transformation, operational research, and professional development in logistics and supply chain management.
Leave a Reply