Harnessing Social Network Analysis to Elevate Product Quality in Complex Engineering

When people hear the phrase "social network", they usually think of Facebook or Twitter. Yet this concept has sparked a wealth of academic research.

Engineers who build complex systems like cars and aircraft have long worked within their own communication networks. As with any network, gaps arise. Operations specialist Wallace Hopp and colleagues recently identified these gaps as "coordination deficits," which can drive costly errors if left unaddressed.

Hopp, Herrick Professor of Manufacturing at Ross, reports that, according to industry managers, roughly 60% of quality issues stem from manufacturing and 40% from design. About 20% of design faults arise from poor communication, indicating that coordination problems cause as many errors as individual mistakes—an eye‑opening insight.

Hopp co‑authored the study The Impact of Misalignment of Organization Structure and Product Architecture on Quality in Complex Product Development with Bilal Gokpinar (University College London) and Seyed Iravani (Northwestern University).

The research maps and quantifies communication breakdowns within a major automaker’s design network. By mining engineering change orders, the team built a social network that yields a statistically robust model. This tool helps managers coordinate projects across global engineering hubs and can be adapted by firms seeking to predict and prevent quality issues in complex products.

The practical takeaway is stark: a 2006 Bloomberg BusinessWeek and BCG survey found that senior leaders rank coordination failures as the second‑largest obstacle to innovation.

A Blessing in Disguise

While previous work has mapped product‑development networks via surveys, Hopp and colleagues realized this method was impractical for their study. They needed data from vehicles that had been on the market for at least a year to access warranty claims, meaning the design phase dated back up to five years. With thousands of engineers involved, many had moved or departed, making recall unreliable even if the right people were surveyed.

Hopp proposed analyzing email and phone logs, but the firm rejected the idea as overly invasive. Instead, the team turned to engineering change orders—comprehensive records of every design step. From these, they extracted a detailed map of vehicle systems, the engineers responsible, and their interaction frequencies.

Using the data, Hopp’s team ranked vehicle systems by complexity and quantified the attention each system received, based on the communication volume among its engineers.

Hopp notes, "To our knowledge, this is the first instance of building a social network from an archival engineering database."

The analysis highlighted systems with coordination deficits—gaps between a system’s complexity and its communication levels—which correlated positively with defects reported in warranty claims.

Medium‑complexity systems exhibited the greatest coordination deficits and defect risk, as highly complex systems naturally attract extensive collaboration, while simple ones require minimal coordination. The intermediate tier often slips through the cracks.

"To uncover misalignments between an organization and its product, quantify the mismatch between their networks," Hopp advises. "We’ve developed a straightforward mathematical framework that companies can readily apply."

The automaker has implemented an interactive org chart to enhance engineer communication and coordination. By providing a holistic project view, engineers can quickly identify appropriate collaborators. Hopp plans to study the tool’s effect on organizational‑product alignment.

Going Global

The team also examined how to steer global design efforts. They discovered that for complex subsystems spread across multiple sites, the sheer number of locations predicts design delays—an effect especially pronounced for highly interconnected subsystems.

Balancing global expertise is challenging; the optimal electrical‑systems team may reside in Europe, even if the bulk of design occurs in North America.

“You need not centralize everything, but reducing design sites for highly central systems yields better results,” Hopp explains. “Shifting a few key personnel to cut the number of locations from four to three—or from three to two—can slash delays. Such analysis will grow in importance as firms globalize their design teams.”

For Hopp, leveraging engineering change orders for network analysis is promising, as the discipline of network science expands.

“The network lens is here to stay and meshes seamlessly with engineering change‑order systems,” Hopp asserts. “Complex products generate vast amounts of relevant data, yet we only now recognize its value. It’s a hidden gold mine.”

Figure 1: Superposition of product and organization networks. When connectivity between subsystems in the product architecture network exceeds connectivity in the organizational coordination network, the resulting "coordination deficit" can lead to design flaws that cause quality problems and, hence, warranty claims.