A medical device manufacturer recorded a 2.3 percent defect rate on a catheter assembly. Corrective actions fired, and the rate fell to 1.1 percent. For the next eight years, every quality review, every supplier scorecard, and every management meeting used that original 2.3 percent as the baseline reference. The team celebrated cutting defects by more than half. Management blocked further capital investments that could have driven the rate below 0.3 percent, satisfied that 1.1 percent looked excellent against the 2.3 percent anchor.
Competitors never had a 2.3 percent incident to anchor against. They compared themselves to zero, not to a historical disaster. By the time the manufacturer lost two major contracts to competitors running at 0.2 percent defect rates, the anchor had done its damage. The reference point that once made leadership feel secure had become the blind spot that cost them their market position.
This is the anchoring effect. It is a cognitive bias where the brain latches onto the first piece of information encountered and adjusts insufficiently from there. In manufacturing environments saturated with targets and historical data, anchoring systematically limits ambition. Organisations optimise against their own past instead of against competitive reality, locking mediocrity into their quality management systems.
The Mechanics of Cognitive Anchoring
First documented by Tversky and Kahneman, the anchoring effect describes the human tendency to rely too heavily on an initial piece of data when making subsequent judgments. When you encounter a number, your brain uses it as a baseline and adjusts outward. The adjustment is almost always insufficient. The final judgment remains contaminated by the initial value, even when superior data is available.
In a factory, numbers dictate behaviour. OEE, Cpk, PPM, and audit scores frame daily decisions. When a historical data point enters the system, it establishes a reference point that resists displacement. If a line starts up and runs at 85 percent first-pass yield, subsequent improvements to 92 percent feel like a triumph. The 85 percent anchor makes 92 percent look like success, concealing the reality that the industry benchmark for that process is 96 percent.
The bias persists because historical baselines feel rational. Comparing current performance to past performance seems like standard continuous improvement. But this internal comparison anchors ambition to history rather than to potential. Organisations that benchmark exclusively against their own past almost always improve more slowly than those that measure themselves against best-in-class performance or theoretical limits.
How Anchor Bias Infects Quality Targets
Anchoring in quality management rarely announces itself. It hides inside standard engineering practices and established KPI structures. A manufacturing dimension carries a specification of 10.0 mm plus or minus 0.5 mm. The process runs at 10.4 mm. When challenged on why the process is not centred, the engineering team responds that the dimension is within specification. The specification limit has become the anchor.
The organisation stops asking what the optimal value is and instead asks how close the process can run to the boundary without triggering a nonconformance. The nominal dimension, which represents the design intent, becomes an afterthought. The entire quality strategy shifts from optimisation to boundary management. The result is a process that technically meets requirements but produces functionally inferior product compared to a competitor who targets the nominal.
Audit scores fall victim to the same mechanism. An automotive supplier receives an IATF 16949 customer audit score of 82. The next year they score 87, then 89. Internal teams celebrate the upward trajectory. But the customer's threshold for preferred supplier status is 92. The organisation anchored its emotional response to incremental improvement rather than to the absolute threshold that actually governs the business relationship.
Sample sizes create another layer of anchor distortion. A quality engineer tests 30 parts from a 50,000-unit lot and finds two defects, generating a reported rate of 6.7 percent. When a later sample of 300 parts finds 11 defects, yielding a rate of 3.7 percent, the team reports significant improvement. The reality is that the initial 6.7 percent figure carried a wide confidence interval. Comparing an imprecise early estimate to a more precise later measurement creates the illusion of progress where none exists.
The Organisational Politics of Anchors
Anchors survive because they serve organisational politics. A quality manager who reduced defects from 4.2 percent to 2.8 percent has a compelling narrative for their performance review. Reducing defects by a third sounds like a story worth telling. Acknowledging that the 2.8 percent figure remains above the industry median of 1.9 percent presents the same data without the anchor, and it tells a much less flattering story.
People protect their anchors because their anchors protect their narratives. Once a number enters a KPI dashboard, a supplier scorecard, or a PPAP submission, it takes on a life of its own. It gets referenced in quarterly reports and embedded in strategic plans. It becomes part of the organisational vocabulary. Removing it requires changing how people think and talk about performance, which is harder than updating a spreadsheet.
A surprising number of manufacturing standards started as a rough guess in a meeting fifteen years ago.
The ambiguity of quality metrics makes this worse. Whether a 2 percent defect rate is acceptable depends on the product, the industry, and the cost of failure. Facing this ambiguity, the brain gravitates toward whatever reference point is immediately available. In the absence of deliberate external benchmarking, the available reference point is always the internal historical number.
Anatomy of an Anchoring Failure
An aerospace supplier develops a new turbine blade casting process. During AS9100 qualification, the process produces an 8 percent scrap rate. The qualification team expected first-article challenges and documents the 8 percent figure as expected initial performance in their report. The process transitions to production.
Over six months, the scrap rate drops to 5 percent. The quality team celebrates the trend and management approves further capital investment. A year later, the rate plateaus at 4.2 percent. Progress has slowed, but the team remains satisfied because they are still improving relative to that original 8 percent anchor documented during qualification.
Internal vs External Benchmarking
What anchored teams do
- Compare current scrap rate to the initial qualification rate
- Celebrate movement away from an internal historical baseline
- Accept plateaus as long as performance beats the starting point
- Build cost models assuming the plateaued rate is standard
What benchmark-driven teams do
- Compare current scrap rate to theoretical process limits
- Evaluate improvement velocity against industry competitors
- Trigger deep-dive 8D investigations when progress flatlines
- Build cost models targeting zero-defect theoretical optima
A competitor developed a similar process from scratch using a different methodology. They established their baseline against the theoretical limit of the equipment, not against an initial development struggle. Their production scrap rate sits at 1.5 percent. The original supplier loses the next contract renewal because their pricing, built on the assumption that 4 percent scrap was acceptable, cannot compete with a supplier whose scrap costs are less than half as high.
Breaking the Anchor: External Reference Points
The most effective countermeasure against anchoring is relentless external comparison. Stop measuring current performance exclusively against past performance. Measure it against industry benchmarks, competitor data, theoretical limits, and customer expectations. If your industry median defect rate is 1.2 percent and you operate at 2.8 percent, the fact that you were at 4.2 percent last year does not change your competitive position. You are still behind.
Establish a practice of benchmarking against three external reference points simultaneously: industry average, best-in-class, and theoretical limit. Present these alongside historical trends in every management review. If a quality dashboard shows only current versus prior year, it is an anchoring trap. A dashboard that shows current, prior year, industry median, best-in-class, and target gives the brain multiple reference points, partially cancelling the distortion of any single number.
Baseline Reset Protocol for Quality Reviews
- 01Acknowledge historical dataDocument the current baseline and the improvement trajectory achieved over the prior period.
- 02Extract external benchmarksPull competitor data, customer requirements, and theoretical process limits for the same metrics.
- 03Discard the internal anchorExplicitly remove the historical baseline from the target-setting discussion and documentation.
- 04Set targets against potentialDefine the next cycle's goals using only the external benchmarks and equipment capabilities.
Reset baselines periodically. At the start of every major improvement cycle, formally discard historical performance as a reference point. Ask what the targets would be if the organisation started from scratch today. The answer becomes the new anchor. This is uncomfortable because it eliminates the narrative of continuous improvement relative to a disastrous starting point. That discomfort is exactly why the exercise works.
Structural Defences Against Anchor Bias
Separate estimation from evaluation. Anchoring is strongest when the people setting targets are exposed to irrelevant numbers before making their judgment. If the same team that sets annual quality targets also evaluates whether those targets were achieved, the targets themselves become anchors that distort the evaluation. Have targets set by one group based on external benchmarks, and performance evaluated by another.
Interrogate every historical number cited in a quality discussion. Ask where the number came from, how it was calculated, and whether it is still relevant. A shocking number of manufacturing tolerances, scrap allowances, and PPM targets started as rough engineering estimates decades ago. The guess became a target, the target became a specification, and the specification became an untouchable anchor treated as immutable truth.
In my experience auditing plants across automotive and aerospace, I have rarely found a historical quality target that could survive basic scrutiny regarding its origins. The anchoring effect thrives in systems that confuse established practice with optimised practice. When standard work and visual management boards lock an unverified baseline into daily routines, organisations stop searching for the true process capability.
Quality professionals must manage reference points as deliberately as they manage processes. Every chart presented, every baseline referenced, every comparison made sets an anchor in the minds of the audience. Consistently choosing internal historical anchors biases the organisation toward complacency. Consistently choosing external, aspirational anchors biases it toward ambition. The choice of anchor is never neutral. It is a decision with measurable consequences, and it must be made with full awareness of its power.