In 1967, Melvin Conway observed that any organisation designing a system will produce a design whose structure mirrors the organisation's communication structure. Conway was writing about software. He might as well have been writing about your IATF 16949 or AS9100 quality management system.
Your QMS does not look the way it does because of ISO 9001. It does not look the way it does because of your customers' requirements. It looks the way it does because of your organisational chart. The people who wrote your procedures could only coordinate within the communication channels they had. They built processes for the people they talk to every day.
This structural mirroring is a problem because defects do not respect departmental boundaries. They flow through the gaps between silos like electricity through a short circuit. If your communication structure has gaps, your quality system has gaps. The 8D corrective actions you write are usually treating the symptoms of those architectural flaws.
The Architecture of Accidents
Consider a typical manufacturing plant. Engineering designs the product. Purchasing sources the material. Production builds the assembly. Quality inspects the result. Shipping sends it out. Each function has its own manager, its own KPIs, its own departmental meetings, and its own technical vocabulary.
Now look at your QMS documentation. The structure matches exactly. There is a design control process owned by Engineering, a supplier quality process owned by Purchasing, and a production part approval process owned by Manufacturing. Quality owns the final inspection. Nobody writes the procedure for what happens between these domains because nobody communicates outside their own silo long enough to realise the interface exists.
I have audited plants where every department was internally compliant. Engineering had perfect phase-gate reviews. Production had rigorous standard work. Quality had meticulous inspection records. Yet they were shipping leaking assemblies. The defect lived entirely in the space between their documented procedures.
Your quality system is designed to match your communication structure. Wherever communication breaks down between functions, your quality system has a hole. Defects find that hole. They pass through the unmanaged handoffs because the organisation structurally eliminated anyone's responsibility for catching them.
The Handoff Problem in Practice
Take a persistent field failure on a precision-machined housing. The customer issues a quality alert. Warranty claims mount. An 8D investigation begins. The cross-functional team assembles. What they find is not a single failure, but a chain of handoff failures.
Engineering designed a sharp edge transition because the CAD software defaulted to it and the design review focused on dimensional tolerances, not edge conditions. Manufacturing programmed the toolpath for minimum cycle time. The sharp transition naturally generated a burr. The burr was not on the inspection checklist because the drawing did not specify burr limits.
Shipping packed the parts in bulk. Vibration during transit pushed the burr into the critical sealing surface. Every department had executed its documented procedure correctly. Every internal audit had passed. The root cause was not a failure within a function. It was the complete absence of control at the boundaries between functions.
The engineering team never spoke to the machinists. The machinists never reported the burr because it was not a defined defect. Quality inspected to the drawing. The customer received a product designed to fail by the very structure of the organisation that built it.
Mapping the Procedural Mirror
Run this diagnostic on your own QMS. Take your core procedures and map them to your organisational chart. Do not map the chart to the procedures. Start with the documentation. Ask who wrote it, who owns it, who reviews it, and whose daily work it is actually designed to serve.
You will find that each procedure serves the department that authored it. The incoming inspection procedure tells inspectors what to check. It does not specify what Purchasing must verify before the material is even ordered. The CAPA procedure describes form completion and closure sign-offs. It does not mandate the cross-functional root cause analysis required on the shop floor.
The PPAP procedure describes the submission levels and approval signatures. It fails to capture the manufacturing feasibility questions Production needed answered before tooling was cut. The PFMEA is a risk register owned by Engineering. It is rarely updated with the failure modes Production encounters during actual serial builds.
People write procedures for the people they communicate with. The critical interfaces, the moments where information must pass from one technical domain to another, are poorly defined or entirely absent. This is Conway's Law at work.
Departmental Procedures vs Cross-Functional Reality
What siloed procedures dictate
- Incoming inspection lists dimensions for Quality to verify on arrival.
- CAPA forms focus on containment actions and closure timelines.
- PFMEA is authored by Engineering during the APQP launch phase.
- Design reviews measure compliance against drawing standards.
What cross-functional reality requires
- Supplier development requires Purchasing to verify process capability at source.
- CAPA drives a multi-disciplinary root cause investigation on the floor.
- PFMEA is a living document updated by Production based on serial run data.
- Design reviews evaluate downstream manufacturability and assembly constraints.
The Cross-Functional Illusion
Most plants claim to have cross-functional communication. They point to daily production meetings and monthly management reviews. They do not have cross-functional communication. They have parallel monologue. Representatives from different departments sit in the same room and take turns presenting their own functional metrics.
True cross-functional communication means the design engineer understands the constraint of the CNC machine that will cut the part. It means the inspector understands the customer's assembly application well enough to know which dimensional deviations actually matter and which are cosmetic.
It means the purchasing agent understands the downstream welding process well enough to know that sourcing a cheaper alloy with slightly different trace elements will cause a downstream cracking problem that costs ten times the material savings. This depth of shared understanding is never achieved through weekly status meetings. It requires structural integration.
It requires shared accountability, shared metrics, shared workspaces, and shared problems. Most importantly, it requires leadership to recognise that a management review meeting cannot substitute for actual communication pathways.
Departments have budgets, headcount, and management attention. The spaces between them have nothing but a shared spreadsheet.
The Silo Multiplier Effect
There is a mathematical dimension to Conway's Law that directly impacts quality costs. The number of communication channels in an organisation grows factorially with the number of people. However, the number of effective communication channels grows much more slowly, because organisational boundaries actively block them.
Adding departments does not add communication capacity. It subtracts it. Every new departmental boundary creates a new interface that must be managed, a new handoff point where information degrades, and a new environment where defects can hide. An organisation with four functional departments has six pairwise interfaces to manage.
Double the departmental structure to eight functions, and you have twenty-eight interfaces. The complexity does not grow linearly. It grows combinatorially. Each interface is a potential quality gap, and each gap requires dedicated resources that the organisation almost never allocates.
Departments have budgets, headcount, and dedicated management. The interfaces between departments have nothing. Perhaps a standing meeting that everyone dreads. Perhaps a procedure that has not been updated since the last external audit. Your QMS is only as strong as its weakest interface.
Applying the Inverse Conway Maneuver
The technology sector has developed a response to Conway's Law that quality professionals must adopt. It is called the Inverse Conway Maneuver. Instead of letting your organisational structure dictate your QMS architecture, you deliberately restructure the organisation to produce the quality system you actually need.
If you want a quality system that prevents defects rather than merely detecting them, you must redesign the communication pathways. Embed quality engineers directly into production cells rather than isolating them in a centralised inspection lab. Give them shared objectives with the production team. Make them responsible for OEE and delivery, not just defect counts.
When quality engineers and production operators share the same daily experience, the procedures they write together will naturally address the boundaries between their domains. The PFMEA becomes a shop-floor tool rather than a launch-phase compliance document. The control plan reflects actual process variation, not just engineering theory.
Implementing the Inverse Conway Maneuver in Quality
- 01Define target QMS architectureMap the defect prevention flow you need, regardless of the current org chart.
- 02Establish interface ownershipName specific individuals accountable for information transfer at each functional boundary.
- 03Deploy cross-functional metricsMeasure handoff quality and inter-departmental defect origins, not just functional yield.
- 04Execute role rotationsPlace engineers on the production floor and inspectors in customer plants to build lasting channels.
Make handoff points explicit and owned. In software engineering, interfaces between components are defined as strict contracts. In quality engineering, you must define exactly what information must pass between each function. Specify the format, the confidence level required, and the named individual accountable for the transfer.
Rotate people across these boundaries. Nothing breaks down organisational silos faster than personnel who have worked on both sides of them. When an engineer has spent six months running the production line, they design for manufacturability. When an inspector has visited the customer's assembly plant, they evaluate deviations differently.
These rotations create personal communication channels that persist long after the assignment ends. Those channels become the pathways through which your quality system naturally integrates. The organisation stops mirroring its silos and starts mirroring its actual value stream.
Measuring the Gaps, Not Just the Functions
Standard quality metrics measure departmental performance in isolation. First-pass yield, scrap rate, on-time delivery, and customer PPM are all critical indicators. They are also fundamentally insufficient. They tell you how well a department performed in a vacuum. They tell you nothing about the friction between departments.
If you want to measure the real health of your QMS, you must measure the interfaces. Track how many defects originated at a functional boundary. Quantify how many corrective actions trace back to a communication failure between Engineering and Production, or between Purchasing and Manufacturing.
Measure the time spent on rework caused by incomplete information transfer. Track the number of engineering change orders that require rework because manufacturing feasibility was not reviewed. These interface metrics will reveal the structural weaknesses that traditional quality scorecards completely miss.
The customer does not experience your departments. They do not experience your engineering process, your manufacturing process, and your inspection process as separate events. They experience a single product. The customer experiences an integrated system. Your task is to build one. You cannot build an integrated product with a disintegrated organisation.
Conway's Law is not a law of physics. It is a strong pattern, and one that can be broken with deliberate, structural effort. The organisations that build resilient quality systems are the ones that recognised the mirror effect and chose to shatter it. They made communication explicit. They assigned ownership. They measured the gaps. Your org chart is not your quality system. But if you do not actively restructure it, it will become one.