The quality improvement project scoped for three months enters its ninth. The CAPA flagged as urgent sits open for 472 days. The validation protocol agreed as straightforward requires three times the runs anyone anticipated. These are not exceptional failures. They are the standard operating mode of quality management.
Every post-mortem produces the same refrain: we should have known. Yet the next project receives the same optimistic timeline, the next budget the same rounding down, the next CAPA the same aggressive compression. The cycle persists because the Planning Fallacy, formally described by Daniel Kahneman and Amos Tversky, is the most deeply embedded cognitive bias in organisational life.
Kahneman himself admitted falling victim to it while writing the very chapter that described the phenomenon. His team estimated one to two years; the work took eight. They had data from comparable projects and still believed they would be different. The bias does not yield to evidence.
In quality management, the consequences extend beyond missed milestones. When timelines collapse, organisations do not extend the work. They compress it. They reduce validation runs, skip edge cases, and close CAPAs before effectiveness checks complete. Every shortcut becomes a latent defect. I have spent two decades auditing systems across automotive and aerospace, and I rank systematic underestimation as the primary driver of escaped defects I have investigated.
The Anatomy of an Underestimate
A planning failure is never a single bad estimate. It is a cascade of compounding underestimates, each one small enough to seem reasonable in isolation but devastating in aggregate. Understanding the cascade is the first step toward interrupting it, because each stage operates on a different failure mechanism.
First comes the scope underestimate. A team investigating weld defects identifies five key parameters and builds a plan around them. The actual number of variables affecting weld quality turns out to be fourteen. The scope has grown by nearly 200 percent, but the timeline has not moved. The PFMEA is updated after the fact, if at all.
Second is the dependency underestimate. The project requires input from the materials lab, but the lab is backlogged with customer complaints. It requires a work instruction change, but that change needs review by three departments that have not met in two months. It requires new measurement equipment, and procurement takes 90 days. Each dependency is a delay multiplier, and dependency chains are consistently longer than anyone plans for.
Third is the rework underestimate. The team assumes their first approach will work. It does not. The gauge R&R fails and the measurement system requires redesign. The process capability study shows Cpk well below 1.33 and the team returns to the drawing board. The 8D corrective action does not hold and the defect returns. Every iteration adds time that was never in the plan.
Fourth is organisational friction. Approvals take longer than expected. Key people are unavailable. Priorities shift. Budgets get reallocated. The project champion leaves. These are not exceptional events. They are the baseline operating conditions of every manufacturing organisation, yet they are systematically excluded from project plans as if wishing could eliminate them.
Why Quality Work Is Especially Vulnerable
Every engineering discipline deals with the Planning Fallacy, but quality management is uniquely susceptible. The structural reasons are specific and they compound. A CAPA might originate in a customer complaint, but its root cause lies in engineering, its corrective action in manufacturing, its verification in the quality lab, and its effectiveness check in field performance. Each handoff is a potential delay, and the number of handoffs is almost always underestimated.
Quality work deals with variability and anomaly. When you plan a production run, you have years of data on cycle times, downtime, and yields. When you plan a quality investigation, you are dealing with something that has not been adequately characterised. The uncertainty is fundamentally higher, and the Planning Fallacy exploits uncertainty by allowing planners to fill gaps with optimistic assumptions rather than worst-case scenarios.
Regulatory scrutiny adds another layer. A validation protocol that would be straightforward in an unregulated environment becomes a months-long process when every deviation must be documented under FDA, EASA, or IATF 16949 requirements, every change justified, and every conclusion defensible to an auditor. Quality professionals who have survived multiple audits still underestimate how much documentation and review add to a project timeline.
Quality work is also perceived as overhead. In organisations where quality is a cost center rather than a value driver, quality projects compete for resources against production and commercial priorities. When resources tighten, quality projects get deferred. The deferral is never factored into the original timeline because the deferral is politically difficult to name during the planning phase.
The Mechanics of Timeline Compression
When a quality project runs over schedule, the organisation faces a binary choice: extend the timeline or compress the work. Most choose compression. The mechanism is rarely a formal decision. It is a series of small, silent compromises made by people under pressure who know the deadline is immovable but cannot get the resources to meet it honestly.
Compression takes predictable forms. Teams reduce the number of validation runs from the statistically valid sample to what the schedule allows. They skip edge cases in the investigation because the mean looks acceptable. They close the CAPA before the effectiveness check is complete because the auditor is coming and the metric must improve. They accept good enough evidence when the standard demands thorough evidence.
How timeline pressure reshapes quality work
What was planned
- Full PPAP submission with all 18 elements verified
- Validation runs at statistically valid sample sizes with defined acceptance criteria
- 8D with root cause confirmed by two independent methods before closure
- Effectiveness check conducted over two full production cycles after implementation
What compression produces
- PPAP elements submitted with production trial run data that was never repeated under serial conditions
- Sample sizes reduced to what the remaining timeline allows, undermining Cpk confidence
- 8D closed after first plausible root cause, no verification of recurrence prevention
- Effectiveness check shortened to one shift, or deferred indefinitely as a follow-up action
The Planning Fallacy also erodes organisational trust in a specific pattern. When quality teams consistently miss deadlines, leadership stops believing their estimates. The natural response is to impose more aggressive timelines, which worsens the problem. The quality team knows the timeline is unrealistic but has stopped fighting because resistance is futile. They nod, commit to the impossible schedule, and silently prioritise what they can accomplish.
This is how organisations develop a shadow quality system: the one documented in procedures and the one operating on the floor. The gap between them is where defects live. The gap is created not by malice or incompetence but by timelines that were never achievable, enforced by managers who needed a number to report upward, accepted by engineers who had no mechanism to push back.
Reference Class Forecasting: The Outside View
Kahneman proposed a specific antidote to the Planning Fallacy: reference class forecasting. Instead of planning a project from the inside out, building up a timeline task by task with optimistic estimates, plan it from the outside in. Look at similar completed projects and use their actual duration, cost, and outcomes as the baseline for the new estimate.
In quality management, this means maintaining a database of past projects with their original estimates and actual results. How long did the last five CAPAs of comparable complexity take to close? How much did the last three equipment validations cost? How many runs did the last process qualification actually require before achieving the target capability? This data exists in every quality organisation's QMS records, but it is almost never used for planning.
The gap between the inside view and the outside view is not a difference of opinion. It is the difference between hope and evidence.
When a quality team says the new CAPA will take 30 days, the reference class says the last 20 similar CAPAs took an average of 90 days. The inside view says this one is different, the root cause is obvious, the team has learned. The outside view says they said that last time too. Implementing reference class forecasting requires discipline: tracking not just whether a CAPA was closed but how long it took, how many iterations were required, and what obstacles appeared.
It means creating categories of quality projects, simple CAPAs versus complex CAPAs, process validations versus system implementations, and maintaining actual performance data for each. And it means having the professional courage to present the outside-view estimate to leadership, even when it is not the number they want to hear. In my experience building greenfield QA functions for automotive plants with over 900 employees, the data always wins the argument if you bring it early enough.
Structural Defenses Against Underestimation
Beyond reference class forecasting, quality organisations can build structural defences that reduce the impact of the Planning Fallacy. These are not cultural changes or mindset shifts. They are concrete mechanisms that change how estimates are produced, reviewed, and tracked.
Calibrated estimation workflow for quality projects
- 01Reference class pullRetrieve actual duration and cost data from the last 5 to 10 comparable projects before writing any new estimate.
- 02Pre-mortemAsk the team to imagine the project has failed. List every failure mode identified and build the response into the plan.
- 03Buffer allocationAdd 25 to 50 percent of estimated duration as an explicit, visible buffer for unknowns. Do not hide it inside task estimates.
- 04Estimate-target separationPresent the evidence-based estimate and the leadership target as two distinct numbers. Make the gap explicit and the bridge funded.
- 05Post-project calibrationRecord actuals against estimates. If the team is consistently 40 percent low, apply the correction factor to the next project.
Explicit buffers are the simplest defence and the rarest in practice. Every quality project should include a buffer of 25 to 50 percent of estimated duration, allocated specifically for the unknowns the Planning Fallacy will inevitably reveal. This is not padding. Padding is hidden inflation added because nobody trusts the number. A buffer is a visible, justified allocation based on historical variance. The difference is transparency.
Pre-mortems are the second mechanism. Before a quality project begins, gather the team and ask what went wrong if this project has failed in six months. The exercise forces the team to articulate risks they have been suppressing and build responses into the plan. The pre-mortem works because it shifts the cognitive frame from optimism to threat assessment, and threat assessment is where quality professionals perform best.
The third mechanism is separating estimates from targets. An estimate is what you believe will happen based on evidence. A target is what you want to happen based on ambition. In most quality organisations these are the same number, and that conflation is the root of the problem. The estimate should be derived from the reference class. The target can be more aggressive, but only if the organisation funds the additional resources and accepts the additional risk required to close the gap.
The fourth is tracking estimate accuracy over time. If your quality team's estimates are consistently 40 percent low, the correction is mechanical: add 40 percent to the next estimate. This is not pessimism. It is calibration. At a major aerospace manufacturer, I introduced routing verification KPIs that achieved a 97 percent reduction in internal lead time, and the foundation was not better intention. It was measurement followed by calibration.
Making the Cost of Shortcuts Visible
When a project runs late and compression is on the table, the decision is almost always made without quantifying the risk. The engineering team reduces validation runs. The quality manager agrees because the alternative is missing a delivery commitment. Nobody calculates the probability that the skipped runs would have detected a problem, or the cost of that problem escaping to the field.
Making the cost of shortcuts visible does not guarantee better decisions. It prevents the organisation from making them blindly. When the team reduces a validation sample size from 30 parts to 10, someone should document the statistical consequence: the confidence interval widens, the probability of detecting a shift decreases, and the residual risk is transferred from the project schedule to the customer.
This documentation serves two purposes. It forces a deliberate decision rather than a silent drift. And it creates an audit trail that links the shortcut to its consequence when the defect eventually escapes. Organisations that make shortcut costs visible tend to take fewer shortcuts, not because they become more virtuous but because the trade-off becomes explicit and defensible decision-makers must own it.
The Planning Fallacy is wired into human cognition. It will not be eliminated by training or awareness. But in quality management, where the consequences of underestimation are measured in defects, recalls, and customer safety, the obligation to resist it through structural mechanism is not aspirational. It is professional. The organisations that get quality right are not the ones with the most sophisticated tools. They are the ones that plan honestly, budget for friction, and defend evidence-based estimates against pressure. That discipline is the difference between a quality system that works and one that merely exists on paper.