Flying High Again - Predict Supply Chain Defects with Analytics

How a major aerospace and defense contractor solved its quality issues by using advanced analytics to predict when and where defects occur in the supply chain

Second in a series of two articles on achieving multi-tier visibility into your supply chain

When supply chain professionals tap open their Twitter feeds, many undoubtedly thank their lucky stars that they weren’t involved in the latest failures that are flying across their screen. Can you imagine being the CPO of the aerospace company who missed production deadlines by years due to one small part? You can’t blame them for cringing when they power on their smart phones. This is the stuff nightmares are made of.

Instead of simply scrolling with a grateful sigh of relief, though, forward-thinking procurement chiefs and their teams can activate data and analytics to achieve critical insights deep within their supply chain (known as “Tier N” visibility) that can prevent disaster.

In our previous article, we described why Tier N visibility is so critical and how analytics is helping companies overcome the challenges associated with traditional supplier surveying to deliver transparency that’s faster, broader and deeper than ever before.

In this installment, we examine the potential ramifications of missing even the smallest issues in your supply chain, and we offer the story of how one aerospace leader has gotten ahead of such challenges.


In the aerospace industry, the margin of error is tiny – one defective part per million is a problem. Imagine if fighter jets began falling out of the sky because of a defective screw. Or, if guided missiles flew off-course into a village because of a faulty sensor.

Given this demand for near-perfection, Dun & Bradstreet worked with a major aerospace company to explore how they could predict and prevent supply disruption. The company felt it had excellent understanding of its Tier 1 suppliers but zero visibility into its sub-tier suppliers. They were very good at reacting to problems after they occurred, and then resolving them. But imagine the opportunity of re-aligning even just a portion of those resources to proactively avoid the issues in the first place. The challenge, then: How could they use data and analytics to find issues with sub-tier suppliers to enable them to foresee pending breakdowns within Tier 1? They wanted the procurement-world equivalent of a tornado warning.

Translating Visibility to Foresight

To begin the analysis, the aerospace company gave Dun & Bradstreet a file of 360 Tier 1 suppliers that could be defined as “direct critical production” – suppliers in areas where a breakdown would be a serious problem.

The suppliers weren’t openly identified as such, but a third of the file was made up of excellent suppliers, a third were moderate, and a third were low performers. All of them had had at least one issue – a “quality escape” where they exceeded the industry-acceptable limit of one defective part per million – but the good ones had had very minor issues, while the low performers were relatively riddled with them.

A team of Dun & Bradstreet data scientists grabbed markers, a whiteboard and D&B’s database of more than hundreds of millions of companies to determine riskiness of each Tier 1’s supply chain, taking into consideration the tier level and the magnitude of risk of each sub-tier supplier. To ensure a fair assessment of all sub-tier suppliers and to eliminate any anomalies – such as a random one-time event that would briefly trip the sub-tier supplier into a higher-risk level – the analysis of each Tier 1’s supply chain incorporated two years of historical data.

They determined that a problem with a Tier 1 supplier would lag problems within the sub-tier. Said another way, a Tier 1 could be performing well, but unbeknownst to that company, one of its suppliers might be experiencing a financial issue or other adverse business circumstance. A few months later, those issues would manifest themselves in a problem with the Tier 1. The Tier 1 got sick a few months after its own suppliers started sneezing.

To be specific, the analysis showed that 65% of the 360 suppliers in the file the aerospace company provided us had at least one sub-tier supplier that registered the worst possible proprietary risk score from Dun & Bradstreet at least six months in advance of the Tier 1’s quality escape.

After seeing the value of this early warning system, the aerospace company asked for another evaluation: A blind test to see if Dun & Bradstreet could actually identify the worst performing Tier 1 suppliers using only the original file of 360 suppliers – remember, the company didn’t disclose which of the 360 were the most or least reliable – and what it knew about the risk in each supplier’s sub-tier.

By analyzing the Tier 2 suppliers that were the most risky over time, Dun & Bradstreet correlated those with the list of the 360 suppliers and correctly identified each of the top 120 worst performing Tier 1 suppliers from the list.

The Power of Prediction

Analytics, combined with the right data, enable companies to have a timely, comprehensive, and ongoing view of their supply chain risk versus the relatively slow and inefficient supplier survey approach. Think of it as LinkedIn for the supply chain, creating deep understanding of first-, second- and third-degree – and beyond – supplier relationships.

Strong analytics mine that data for the predictive indicators – and, in the case of Dun & Bradstreet, model proprietary predictive risk scores – that best correlate to a future supply disruption.

With this knowledge, the aerospace company could target its inspections process on suppliers the data and analytics pointed to as high-risk. It also could avoid late delivery from suppliers by sourcing alternate suppliers for parts where regular suppliers had high Tier 2 exposure.

By knowing when quality issues will occur before they actually do, companies using Dun & Bradstreet's supply chain management solutions can now be more proactive in mitigating risk to avoid disruptions and to gain competitive advantage.

What would predicting quality issues mean for your company?