Few sentences in the project management literature carry as much quiet weight as the claim that, “by evaluating product and project independently, an organization implements an important project learning mechanism that can lead to improvements in project process and outcomes and, hopefully, lead to consistently successful projects” (Klastorin & Mitchell, 2020, p. 559). On its surface the statement reads almost as a truism, because of course an organization should look at both what it built and how it built it. Yet Klastorin and Mitchell (2020) place this claim at the center of their discussion of how a multiproject organization can “systematically learn to improve delivery and process” (p. 556), and they do so precisely because the practice it describes is, in their words, “frequently overlooked by organizations” and “may be partially responsible for the ongoing poor performance … consistently reported by studies on project performance” (p. 557). The sentence is therefore not a truism at all. It is a diagnosis and a prescription bundled together. The diagnosis is that organizations routinely conflate two very different questions, namely whether they built something valuable and whether they managed the building of it well, and that, by conflating them, they blind themselves to the very information they need to improve. The prescription is that these two questions must be asked and answered separately, deliberately, and repeatedly, so that the answer to one never contaminates or excuses the answer to the other.

This paper explains why independent evaluation of product and project functions as a genuine learning mechanism rather than a bureaucratic ritual, and it grounds that explanation in a single, unusually clear real world example, the Sydney Opera House, a building the authors themselves invoke as the archetype of a project that succeeded spectacularly as a product while failing dismally as a project (Klastorin & Mitchell, 2020). Drawing on the textbook’s treatment of project audits, performance measurement, estimation bias, stage gates, and project closure across Chapters 10 through 12, and supplementing that treatment with two accessible online accounts of the Opera House’s construction and its subsequent cultural and economic life, the paper argues that the value of independent evaluation lies in what it protects an organization from. It protects the organization from the seductive and dangerous inference that a good outcome proves a good process. It preserves the diagnostic signal, meaning the deviations, the estimation biases, and the unplanned work, that would otherwise be washed out by the glow of a celebrated result. And it converts each completed project, successful or not, into a structured deposit of knowledge that the next project can withdraw against.

To understand why the two evaluations must be separated, one must first appreciate how easily they collapse into a single judgment in the ordinary course of organizational life. When a project ends, the question that everyone wants answered, whether the sponsor, the customer, the executive committee, or the public, is simply, “Was it a success?” That question feels unitary. It feels like it should have one answer. But Klastorin and Mitchell (2020) insist that it decomposes into two independent dimensions that need not, and often do not, point in the same direction. The first dimension is product performance, which they define as “the quality and/or value of the project outcomes” (p. 556). The second is project process performance, defined as “the quality of the tasks that went into creation of the product” (p. 556). The product is the thing the world receives, and the process is the sequence of planning and execution activities the organization performed to produce it. The first is evaluated against the value proposition established during project initiation, and the second is evaluated against the baseline of planned time, cost, and scope.

The reason these two dimensions can diverge is that they are governed by different causes and measured against different yardsticks. A product can be magnificent because the underlying idea was visionary, because the market shifted favorably, because the design turned out to be more valuable than anyone anticipated, or simply because the deliverable, however painfully produced, does the job extraordinarily well. None of those causes has anything to do with whether the work was planned accurately, scheduled realistically, estimated without bias, or executed efficiently. Conversely, a process can be a model of discipline, with every task planned, every estimate honest, and every milestone hit, and still yield a product that the market rejects or that fails to deliver its promised value, because the value proposition itself was mistaken. The authors’ own examples make the point vivid from the product side. Marvel Studios could set a payback period of one week as a criterion of product success, and Avengers: Infinity War, grossing an estimated $640 million on its opening weekend against an estimated cost of $321 million, achieved a payback period of roughly a day and a half (Klastorin & Mitchell, 2020). That is an overwhelming product success measured purely against the value proposition, and it says nothing whatsoever about whether the film’s production schedule and budget were well managed.

Because the two dimensions are conceptually distinct, an organization that reports only a single verdict is quietly choosing which dimension to privilege, usually without realizing it is choosing at all. In practice, product success tends to win, because product success is visible, celebrated, and consequential, while process performance is technical, internal, and easy to rationalize after the fact. “Does a successful product vindicate poor project performance?” Klastorin and Mitchell (2020) ask pointedly (p. 559). The intuitive answer, the answer an organization gives when it collapses the two questions into one, is yes, of course, look at what we built. The authors’ answer is an emphatic no, and the whole argument for independent evaluation rests on understanding why that intuitive vindication is a trap.

The first and most fundamental reason independent evaluation is a learning mechanism is that it defeats a confound. In any measurement problem, a confound is a variable that distorts the reading of the thing one actually cares about. When an organization evaluates a project by looking at the product alone, the product’s success becomes a confound that suppresses the signal coming from the process. The result may be so good, so publicly acclaimed, that no one is willing to look closely at how it was produced, and the how is exactly where the learning lives.

Consider what a bad process actually consists of, according to Klastorin and Mitchell (2020). It consists of unplanned tasks appearing as a large percentage of planned tasks, which indicates that the work breakdown structure was incomplete or the risk management plan inadequate. It consists of unplanned hours exceeding the buffer set aside for them, which indicates that the planning team undersized its contingencies. It consists of large signed and absolute deviations between actual and planned time and cost. And it consists of systematic estimation bias, in which actuals run consistently above or below plan, revealing a flawed estimating methodology. Every one of these is a diagnostic finding. Every one of them, if surfaced and analyzed, tells the organization something specific and actionable about how to plan and execute the next project better. And every one of them can be present in full force behind a product that the world regards as a triumph.

If the organization evaluates only the product, all of that diagnostic information is thrown away. Worse, it is thrown away silently, never collected, never examined, and never allowed to speak. The product’s success functions as an anesthetic. It numbs the organization to the pain signals that would otherwise motivate change. Independent evaluation is the mechanism that keeps those pain signals alive. By insisting that project process performance be assessed on its own terms, against its own baseline, regardless of how the product turned out, the organization guarantees that the diagnostic findings survive contact with the celebration. This is why Klastorin and Mitchell (2020) can say that the failure to differentiate the two “may be partially responsible for the ongoing poor performance … consistently reported by studies on project performance” (p. 557). Organizations keep managing projects poorly, in part, because their successes keep hiding the evidence of their poor management.

The second reason independence is a learning mechanism is subtler and concerns not the absence of learning but the presence of mislearning. An organization that judges process by product does not merely fail to learn. It often learns the wrong thing, and a wrong lesson is worse than no lesson because it actively degrades future decisions.

Klastorin and Mitchell (2020) make this point with precision in their discussion of the baseline. “If we evaluate a project’s performance against a flawed baseline, we may draw an incorrect conclusion about the source(s) of the project’s challenges and opportunities for improvement” (p. 560). This is a statement about attribution. When a project runs over budget and behind schedule, there are two possible culprits, and the authors name them explicitly, namely project planning and project execution. If the plan was sound and the team simply failed to execute it, the lesson is about execution discipline, meaning tighter progress control, better resource management, and closer supervision. But if the plan was flawed, with biased estimates, an incomplete work breakdown structure, or unassessed risks, then the overrun is not really an execution failure at all. It is a planning failure masquerading as one, and punishing the execution team for it would be both unjust and useless. The organization would fix the wrong thing and leave the real defect untouched, guaranteeing its recurrence.

Independent evaluation is what makes correct attribution possible. The method the authors describe is to begin the process audit with “the systematic evaluation of various planning performance metrics” (Klastorin & Mitchell, 2020, p. 560), including unplanned tasks, unplanned hours, and actual to planned ratios, precisely so that the organization can determine whether it is looking at a planning problem or an execution problem before it draws any conclusion about how to improve. The exhibits on estimation bias dramatize this clearly. A distribution of actual to planned ratios centered on 1.0 with a tight spread indicates good, unbiased estimating. A wide, low distribution centered on 1.0 indicates unbiased but imprecise estimating, where the fix is to identify the missing modifiers, and the authors’ own example is a “Program Medium Size Java Class” whose true effort depends on data sources, calculation intensity, and interactive versus batch processing (Klastorin & Mitchell, 2020). A distribution shifted to the right of 1.0 reveals systematic underestimation bias, a defect in the estimating methodology itself. These are three completely different diagnoses demanding three completely different remedies, and none of them can be read off the product. They can only be read off an independent examination of the process against a properly understood baseline. Without that independence, the organization is guessing at the cause of its problems, and it will just as often repair a healthy part of its process while leaving the diseased part to fester.

The third reason independence is a learning mechanism is that it forces the creation and preservation of the raw material that learning requires. Learning is not an abstraction. It runs on data, and data must be captured, cleaned, indexed, and archived, or it evaporates. Klastorin and Mitchell (2020) are unusually insistent on this point in their treatment of project closure, observing that “every completed project represents a massive amount of data that should be utilized to improve future project performance” (p. 563). Among the closure activities they prescribe are updating the project database with clean task data such as estimates, revised estimates, and actuals, verifying and finalizing the project repository, preparing a final report documenting product and project performance, conducting a final project audit, and revising project planning and management methods, guidelines, and tools if warranted based on the findings of the project audit (Klastorin & Mitchell, 2020).

Notice the structure of that sequence. The clean task data, meaning the estimates alongside the actuals, is exactly the raw material needed to construct the actual to planned distributions that reveal estimation bias. The final report documents product and project performance separately. The final audit, conducted by an independent team, assesses the process. And the revision of methods and guidelines is the actual moment of learning, the point at which the diagnosis becomes a changed practice. This closing loop only functions if the two dimensions have been kept distinct throughout, because the whole purpose of capturing estimates versus actuals is to evaluate the process independently of the product’s fate. An organization that has decided in advance that a good product means a good project has no reason to preserve its estimate versus actual data with any care, because it has no intention of interrogating it.

The authors even guard against a specific way this can go wrong, insisting that the project team’s final report and the independent audit report should “remain separate” (Klastorin & Mitchell, 2020, p. 564). The reason is that the audit may point out failures of process, individuals, or organizations outside the project team, and if the team itself authored those criticisms, the findings would either damage the team’s members after transition or be dismissed as buck passing or justification of poor project performance. This is independence layered on independence. Not only must product and project be evaluated separately, but the process evaluation must be performed by an independent audit team rather than by the people whose work is under review. The architecture is designed at every level to keep the diagnostic signal honest and credible.

The fourth reason draws the previous three together into the phrase that ends the authors’ claim, namely “consistently successful projects” (Klastorin & Mitchell, 2020, p. 559). The operative word is consistently. A single successful product can be a matter of luck, vision, or favorable circumstance. Consistency cannot. Consistency is the signature of a process that reliably converts inputs into valuable outputs, and a reliable process is precisely the thing that independent evaluation builds, because it is the only dimension of the two that is fully within the organization’s repeatable control.

Product success depends heavily on factors an organization cannot summon at will, including the brilliance of a design, the receptiveness of a market, and the alignment of timing. Process quality, by contrast, is an organizational capability that can be measured, diagnosed, and deliberately improved from one project to the next. If an organization wants not one triumph but a portfolio of them, it must invest in the dimension it can actually compound, which is the process, and it can only invest intelligently in what it measures honestly. Independent evaluation is what makes honest measurement of the process possible, because it strips away the vindication reflex, the reflexive appeal to how well it turned out that would otherwise excuse every process defect the moment the product succeeded. Break that reflex and the organization can, project after project, tighten its estimating, complete its work breakdown structure, size its buffers correctly, and manage its risks, accumulating a process capability that makes the next success less a matter of luck and more a matter of design. That accumulation, and only that accumulation, is what turns occasional success into consistent success. The learning mechanism is not merely retrospective bookkeeping; it is the compounding engine by which a project organization gets systematically better at its own core activity.

No case illustrates the necessity and the payoff of independent evaluation more cleanly than the Sydney Opera House, and it is no accident that Klastorin and Mitchell (2020) reach for exactly this example to introduce the very sentence this paper is built around. “The Sydney Opera house is a marvelous venue visited by more than 8.2 million people every year,” they observe, and “yet the time required to complete the project to build the facility required 350% of the time originally estimated and the final cost was an astonishing 1,450% of the original cost estimate. By any measure of product success, the Opera House has been successful. By any measure of project success (e.g., adherence to schedule, budget), the opera house was a dismal failure” (p. 559). The building is, in effect, a permanent monument to the divergence of the two dimensions, a structure whose sails announce a product triumph while its ledgers record a project catastrophe.

Evaluated purely as a product, meaning as a deliverable measured against the value it creates, the Sydney Opera House is one of the most successful projects of the twentieth century, and its success has, if anything, grown with time. In 2007 it was inscribed as a UNESCO World Heritage Site, an honor that places it among the small set of human works judged to have outstanding universal value (We Build Value, n.d.). Its economic and cultural contribution is not merely symbolic but quantified. A Deloitte assessment cited in that same account valued the building’s social contribution at roughly $11.4 billion in 2023, a 38 percent increase over the prior decade, and estimated that it contributes on the order of $1.2 billion annually to the state economy, while hosting more than 1,800 performances a year and drawing on the order of 1.4 million spectators (We Build Value, n.d.). Even the analyses most concerned with its managerial failings acknowledge that the finished building became a beloved global icon, a product success that is not seriously disputed (UM Planners, n.d.). The building that the press once condemned as an architectural folly was, upon its 1973 inauguration, transformed into a masterpiece, and by every product oriented measure the value proposition of a world defining cultural venue was not merely met but wildly exceeded (We Build Value, n.d.).

Framed in the authors’ own vocabulary, the Opera House’s product performance lies not merely above the lower bound of any reasonable confidence interval established at initiation, but far above the upper bound, the region Klastorin and Mitchell (2020) reserve for outcomes that exceeded the organization’s original expectations.

Evaluated as a project, meaning as a managed process measured against its baseline of planned time and cost, the same undertaking is a textbook disaster, and the online accounts sharpen the picture the textbook sketches. The UM Planners analysis reports an original cost estimate of roughly $7 million against a final cost of well over $100 million, an overrun on the order of 1,300 percent, and an original timeline of four years against an actual construction span of fourteen years, from 1959 to 1973, a full decade beyond schedule (UM Planners, n.d.). The We Build Value account confirms the same fourteen year duration and puts the final cost at $102 million against the $7 million estimate (We Build Value, n.d.). The precise multiples vary by source and by how one defines the start and end points, and the textbook figures of 350 percent of estimated time and 1,450 percent of estimated cost sit within the same range of catastrophe (Klastorin & Mitchell, 2020), but the direction and magnitude are unmistakable across every account. The process by which this magnificent product was created blew through its schedule by a factor of several and its budget by more than an order of magnitude.

The causes of that process failure are, importantly, the very causes Klastorin and Mitchell (2020) catalog when they explain how to audit planning performance. Construction began before the design was resolved, which is to say before the scope was defined and the work could be honestly estimated, a guarantee of unplanned tasks and unplanned hours on a monumental scale. The relationship between the client and the designer deteriorated to the point that the architect, Jorn Utzon, resigned in 1966, before the building was complete (We Build Value, n.d.). Political pressure to begin visibly, before the plans were ready, converted an ambitious design into an open ended improvisation. In the authors’ terms, this was overwhelmingly a planning failure rather than an execution failure, because the baseline against which the project was measured was a fiction from the start, an estimate produced before anyone could have known what the building would require. The $7 million figure was not a target that disciplined execution could have hit. It was a number attached to a design that did not yet exist.

Now suppose an organization evaluated the Sydney Opera House the way Klastorin and Mitchell (2020) warn against, by letting the product verdict stand for the project verdict. The reasoning would be irresistible and disastrous. It would run as follows: we produced a UNESCO World Heritage Site that generates over a billion dollars a year and is beloved around the world, therefore the project was a success, therefore our approach to it was sound. Every lesson the Opera House has to teach would be lost in that single inference. The organization would conclude that beginning construction before finalizing the design is acceptable, because look how it turned out. It would conclude that its estimating methodology is fine, because the building got built and the world loves it. It would, in short, treat a $7 million estimate that became a $102 million actual as vindicated by ticket sales, and it would carry every one of those defective practices, unexamined, into its next project.

Independent evaluation refuses that inference. By assessing the project process against its own baseline, separately from and indifferent to the product’s acclaim, the organization is forced to record that the estimate was off by more than a factor of ten, that scope was undefined at the start, that unplanned work dominated the effort, and that governance broke down catastrophically. These findings are not diminished one iota by the building’s later glory, because the evaluation was never allowed to reference that glory in the first place. And each finding maps directly onto an actionable improvement of exactly the kind Klastorin and Mitchell (2020) describe an audit producing, namely to finalize design and scope before committing to a cost baseline, to build estimating methods that account for genuine uncertainty rather than political optimism, and to establish governance that can absorb client and designer conflict without losing the designer. This is the learning mechanism operating in full, with the product success and the process failure held apart, so that the success is enjoyed without excusing the failure and the failure is analyzed without tarnishing the success.

The Opera House is not an isolated curiosity. It is the visible tip of a systematic pattern, and that pattern is itself an argument for independent evaluation. As one project management analysis puts it, the building “remains a prime example of how lack of foresight in scheduling and planning can lead to massive delays and cost overruns, even for projects that eventually become world-renowned” (UM Planners, n.d.). That final clause is the whole problem in miniature. Large, complex projects routinely overrun their schedules and budgets and yet deliver products the world ultimately celebrates, from bridges and stadiums to landmark buildings, and when the celebration is allowed to stand in for the verdict, the process failures are never reckoned with. They therefore recur with almost mechanical reliability, project after project, decade after decade. This is chronic mislearning at the scale of an entire sector, the empirical shadow of the authors’ warning about failing to evaluate the two dimensions independently (Klastorin & Mitchell, 2020). The Sydney Opera House shows one organization’s version of the trap, and the persistence of the pattern across the industry shows that escaping it requires exactly the deliberate discipline the authors prescribe.

To call independent evaluation a mechanism is to claim that it is built out of concrete, repeatable organizational practices rather than good intentions, and Klastorin and Mitchell (2020) are careful to specify what those practices are. Understanding them shows how the abstract principle becomes an actual engine of improvement.

The foundational practice is the project audit, which the authors define as “a thorough evaluation of a project’s status and delivery to date” conducted so as to provide “an evaluation of project performance independent of the project team” (Klastorin & Mitchell, 2020, p. 557). The independence is doubly important, because it separates the process evaluation from the product’s fate, and it separates the evaluators from the evaluated. An audit assesses the quality and extent of deliverables, how well results deliver against the value proposition, how actual resource utilization correlates to projected usage, timeliness against schedule, and how well governance and infrastructure are facilitating the work. Crucially, the authors charge the audit team with raising concerns that a project manager may “lack the political capital or influence” to raise (Klastorin & Mitchell, 2020, p. 557), for instance whether management is genuinely supporting the project and providing resources on time. The audit is thus not only a diagnostic instrument but a channel through which uncomfortable truths can reach the organization. And the authors are candid about the limits of this channel. The Cover Oregon project engaged an outside audit firm whose reports “consistently described a project that was increasingly dysfunctional,” yet key managers and politicians ignored the warnings, which proves that audits, “while important and valuable, are not a guarantee of project success” (Klastorin & Mitchell, 2020, p. 557). Independent evaluation supplies the signal. It cannot force the organization to act on it. That caveat is itself a lesson, and one that only independent evaluation could surface.

The second practice is the stage gate or toll gate structure, in which a project is organized into sequential stages and, at the end of each, an evaluation determines whether it should “proceed, should be recycled …, or should be terminated” (Klastorin & Mitchell, 2020, p. 559). The authors explicitly tie this to independent audits, noting that “the objective recommendations of an independent project audit team can be a valuable input for making gate decisions” (p. 559). Stage gates matter because they move the learning mechanism forward in time. Rather than waiting until closure to discover that a process was flawed, the organization interrogates product and process at each gate, catching a broken baseline or a runaway cost trajectory while there is still time to recycle a stage or terminate the investment. In a multiproject environment where resources are scarcer than the demands of the portfolio, gates also reframe termination as “effective project investment rather than recriminations against unsuccessful project teams” (Klastorin & Mitchell, 2020, p. 559), which protects the cultural conditions under which honest evaluation can survive. An organization that treats every early termination as a punishment will soon find its teams concealing exactly the process problems the mechanism is meant to expose.

The third practice is the disciplined measurement of the two dimensions along separate tracks. Product performance is measured against the value proposition established at initiation and refined through planning and change management, so that, as the authors put it, “the basis for evaluating product success is readily available (and independent of measures of project success)” (Klastorin & Mitchell, 2020, p. 559). Project process performance is measured against the planned baseline using the tools developed earlier in the book, namely variance analysis and earned value analysis from Chapter 11, along with the planning metrics of Chapter 12 such as unplanned tasks, unplanned hours, signed and absolute task deviations, and actual to planned ratio distributions (Klastorin & Mitchell, 2020). The earned value apparatus, comprising the actual cost of work performed, the budgeted cost of work scheduled, the budgeted cost of work performed, and the variances and critical ratios derived from them, is the quantitative machinery that makes process evaluation rigorous rather than impressionistic. These tools are the instruments the mechanism runs on. Independent evaluation without earned value and variance analysis would be an aspiration, but with them it becomes a measurement discipline.

The fourth practice is project closure, the stage at which the mechanism’s output is captured for reuse. Closure verifies completion against acceptance criteria, closes out subcontracts and purchase orders, finalizes the repository, updates the database with clean estimate and actual task data, produces a final report documenting both product and project performance, secures formal acceptance, conducts the final and most comprehensive audit, and revises methods and guidelines based on the audit’s findings (Klastorin & Mitchell, 2020). The authors stress that the repository should be “complete, indexed for easy retrieval, and archived” (Klastorin & Mitchell, 2020, p. 563), ideally as a low effort, low cost activity made easy by a document management system established at the project’s start. This is the organizational memory on which all future learning depends. A project whose data is lost at closure teaches nothing to its successors, however much it might have taught in principle, while a project whose estimate versus actual data is preserved and indexed becomes a permanent contributor to the organization’s estimating accuracy. Closure is where the mechanism either compounds or resets to zero.

Underlying all of these practices is the question of who owns them, and here Klastorin and Mitchell (2020) point to project governance, meaning the project management office and the project manager, as the custodians of the learning mechanism. A project management office can provide the independent audit capability, maintain the repository and the estimating database across projects, and enforce the separation of product and process evaluation as a standard rather than an option. The project manager, described in the text as a business leader responsible for delivering the value proposition, is the person who must resist the vindication reflex in the heat of a celebrated delivery and insist that the process be examined on its own terms (Klastorin & Mitchell, 2020). Without governance that institutionalizes independent evaluation, the mechanism depends on the discretion of individuals and will erode the moment attention wanders or a large success arrives to distract everyone. Governance is what makes the mechanism systematic, and to systematically learn to improve delivery and process is the authors’ own description of the goal (Klastorin & Mitchell, 2020).

A fair treatment must acknowledge that independent evaluation is neither free nor sufficient, and Klastorin and Mitchell (2020) are honest about both limits. It is not free, because audits consume time and money, and the authors explicitly instruct that the time and costs of conducting the audit should be included in the plan and estimates for the project, and that the final team members who may be necessary for facilitating the audit be retained through closure (Klastorin & Mitchell, 2020). Evaluation is real work that must be planned and funded like any other. And it is not sufficient, because the Cover Oregon case demonstrates that even excellent, independent, repeatedly delivered audit findings accomplish nothing if the organization’s leadership refuses to act on them. Independent evaluation produces knowledge, but it does not produce the will to use that knowledge.

Yet each of these limits, examined closely, strengthens rather than weakens the case for the mechanism. That evaluation costs money is an argument for planning it deliberately, not for skipping it, and the authors’ response is to build the cost into the estimate, while mature organizations with stable performance can economize by auditing randomly sampled projects rather than every one, an application of statistical process control that presupposes rather than abandons the independent evaluation discipline (Klastorin & Mitchell, 2020). That evaluation is not self executing is an argument for governance that ensures findings reach decision makers with the authority and the incentive to act, which is precisely why the authors locate audit independence, stage gate authority, and project management office ownership at the center of their treatment. The objections define the conditions under which the mechanism works. They do not show that it fails to work. Indeed, the deepest lesson of Cover Oregon is one that only independent evaluation could have produced. The audits were right, the organization was warned, and the failure lay not in the diagnosis but in the response. An organization that had never evaluated the project independently would not even possess that lesson.

There is also a cultural objection worth naming, because it is the one that most often defeats the mechanism in practice. Independent evaluation of the process, especially after a celebrated product, feels ungracious. It feels like looking for problems in a triumph, like refusing to let the team enjoy a win. This is exactly the sentiment the vindication reflex feeds on, and it is why Klastorin and Mitchell (2020) take such care to frame evaluation as “forward looking,” aimed at improving “the organization’s ability to deliver value through projects” (p. 558), and to insulate individuals from the audit’s findings by keeping the audit report separate from the team’s final report and by framing terminations as investment decisions rather than recriminations. The learning mechanism survives only in a culture that can hold two thoughts at once, namely that the organization built something wonderful and that it managed the building of it poorly, and the second thought is not an insult to the first but the price of doing it better next time. The Sydney Opera House earns its place in the textbook precisely because it makes that dual judgment unavoidable. No one can look honestly at that building and its budget and pretend the two verdicts are the same.

The claim by Klastorin and Mitchell (2020) that “by evaluating product and project independently, an organization implements an important project learning mechanism that can lead to improvements in project process and outcomes and, hopefully, lead to consistently successful projects” (p. 559) is true because independence defeats the single most common way organizations fool themselves, which is by letting a good product certify a bad process. When an organization collapses the two dimensions into one verdict, product success, being visible and celebrated, almost always wins, silently suppressing the diagnostic signal that lives in the process. Independent evaluation keeps that signal alive. It preserves the unplanned tasks, the blown buffers, the estimation biases, and the broken baselines that would otherwise be washed out by acclaim. It makes correct attribution possible, so the organization fixes the flaw it actually has rather than the one it imagines, and it forces the capture and preservation of the estimate versus actual data on which all future improvement depends. By breaking the vindication reflex, it lets the organization invest in the one dimension it can actually compound, which is the process, turning occasional, luck dependent success into deliberate, repeatable success.

The Sydney Opera House makes the argument concrete. As a product it is an $11.4 billion cultural asset, a UNESCO World Heritage Site drawing well over a million spectators a year, and as a project it overran its cost estimate by more than an order of magnitude and its schedule by a factor of several, its scope undefined at the start and its architect gone before the end (We Build Value, n.d.; UM Planners, n.d.). An organization that judged the project by the product would learn nothing and would carry every one of those defects into its next undertaking, and the persistence of schedule and cost failures across celebrated megaprojects suggests that this is very nearly what the construction industry as a whole has done (UM Planners, n.d.). An organization that evaluates the two dimensions independently records both truths, enjoys the first without letting it excuse the second, and converts the second into a specific agenda of improvement. That is the learning mechanism, operationalized through independent audits, stage gates, disciplined dual track performance measurement, and rigorous closure, and owned by governance strong enough to make it systematic and to act on what it finds (Klastorin & Mitchell, 2020). It is, in the end, the mechanism by which a project organization refuses to be fooled by its own successes, and, in that refusal, learns to build them on purpose, again and again.

References

Klastorin, T., & Mitchell, G. (2020). Project management: A risk-management approach (1st ed.). Sage.

UM Planners. (n.d.). The Sydney Opera House: A famous example of project schedule failure. https://www.umplanners.com/post/the-sydney-opera-house-a-famous-example-of-schedule-failure

We Build Value. (n.d.). Sydney Opera House: From folly to UNESCO World Heritage Site. https://www.webuildvalue.com/en/infrastructure/sydney-opera-house-story.html