A digital-service team is starting plenty of work, but accepted output is falling while the review queue grows. This lab follows six weeks of evidence to investigate the constraint, distinguish incomplete handoffs from unavailable qualified capacity, and design a two-week improv…
CURRENT EDITION: PMP 2026. Start with Season 2 for the updated certification preparation course. Previous seasons are archived and may not reflect current exam requirements.
The Certified PMP® Audio Course is your complete, on-the-go study companion for mastering the Project Management Professional (PMP®) certification exam. Designed for busy professionals, this Audio Course transforms every aspect of the Project Management Institute’s (PMI) exam content into structured, easy-to-follow lessons that blend clarity with real-world application. Each episode guides you through the principles of project leadership, process integration, stakeholder management, risk, quality, and agile methodologies—ensuring you understand not only the “what” but also the “why” behind each concept. Whether you’re commuting, exercising, or reviewing between meetings, this series helps you study smarter, stay motivated, and move steadily toward exam success.
The PMP® certification is the global gold standard for project managers, validating your ability to lead teams, deliver results, and manage complex projects under real-world constraints. It measures your proficiency across predictive, agile, and hybrid approaches—emphasizing leadership, strategic thinking, and adaptability. Earning the PMP® demonstrates mastery of the three key domains: People, Process, and Business Environment, and signals to employers that you can manage risk, align projects with organizational goals, and deliver value from start to finish.
Developed by BareMetalCyber.com, the Certified PMP® Audio Course provides structured instruction, practical insights, and exam-focused guidance to help you build the confidence, mindset, and discipline needed to pass the PMP® exam and excel as a certified project management professional.
A team can look extremely busy while its customers wait longer every week. Developers begin more items, the board fills with work, and every status conversation includes a promise to catch up. Yet fewer items reach accepted completion. When that happens, the useful question is where work stops moving and what evidence would explain the delay. Continuous improvement starts with that question, then follows through with an authorized change, a measured result, and a decision about what should become part of the way the organization works.
Welcome to Certified, the P M P prepcast. This is episode one hundred one of season one. Today we are exploring how to investigate a delivery problem, test an improvement, and maintain what works.
This lab brings together quality planning, the cost of quality, and continuous improvement. We will follow a digital service through six weeks of evidence, including a two-week improvement experiment. Everything needed to reason through the case will be described aloud. The case is fictional, including its numbers and its authority rules. Its purpose is to practice recognizing what the evidence supports, what remains uncertain, and who can make the next decision. A useful recommendation can acknowledge uncertainty while still helping the team move forward.
The project lasts ten weeks and replaces a paper intake process with a digital service. The team delivers usable slices of the service. Its flow board shows three stages: build, independent review, and accepted completion. The business purpose is reliable intake with less avoidable waiting for users. Counting more started items would not demonstrate that result. Even a higher number of completed items would need to be connected to the service's actual reliability and the waiting experienced by people using it.
There is a precise completion rule. A release item is finished only after it passes its acceptance checks and the required independent review. Independence is part of the agreed control arrangement. The team has authority to adjust local workflow practices, but that authority does not allow it to remove the independent review. Keep this boundary in mind as the queue grows. A faster-looking board obtained by redefining unfinished work as finished would change the measurement and weaken the control instead of demonstrating improvement.
Before interpreting the numbers, establish what each measure counts. Throughput means the number of items accepted during each five-working-day week. The end-of-week review queue counts items whose build work is complete but that are still waiting for review. Median review wait is the middle waiting time, measured in working days from entry into review until review starts, for items reviewed during that week. It is one part of the workflow. It does not measure the full time from starting an item to its accepted completion.
Escaped defects have another important boundary. They are counted during the same five-working-day follow-up window after an item finishes. The report is compiled only after every listed follow-up window has completed. That gives each week's finished work the same opportunity for these defects to be observed. The work belongs to one broadly comparable item class, but detailed complexity can vary. Defect counts still need context because completion volume and work mix differ. A count of defects is not automatically a percentage of defective items.
Here is the first week. Eight items finish. Three items remain in the review queue at the end of the week. Median review wait is two working days, and one escaped defect is recorded within the defined follow-up window. In week two, seven items finish, five remain queued, median review wait rises to three working days, and one escaped defect is recorded. The team is still delivering, but finished output falls while both the queue and waiting time increase.
In week three, six items finish, seven remain queued, median review wait is four working days, and one escaped defect is recorded. In week four, five items finish, nine remain queued, median review wait is five working days, and two escaped defects are recorded. Across these four weeks, finished items move from eight to seven to six to five. The end-of-week queue moves from three to five to seven to nine. Median review wait moves from two to three to four to five working days.
Imagine those measures as separate lines across the same four weeks. Accepted completion slopes downward. Queue size and waiting time slope upward. This is enough to identify an observable delivery problem and investigate the review stage. It is not enough to diagnose the cause by itself. We have four baseline observations, no established statistical control limits, and no evidence that every item is identical. Calling this a statistically proven, out-of-control process would go beyond the information provided. The practical concern is real without that unsupported claim.
A useful problem statement would say that accepted completion declined from eight to five items per week while the review queue grew from three to nine and median review wait increased from two to five working days. Those conditions delay the movement of usable work toward accepted completion. Notice what the statement does not assume. It does not say that developers are careless, that the reviewer is slow, or that independence is unnecessary. The team can now investigate explanations without embedding blame in the definition of the problem.
At the end of week four, the team examines all nine waiting items. Seven already have the evidence required for review. Two lack a required test record, and named owners are preparing those records. The one qualified reviewer has repeatedly been diverted to another project. These details identify at least two contributing conditions: interrupted review capacity and incomplete handoffs for some items. Treating all nine items as unready would ignore the seven that are ready. Treating every delay as a staffing problem would ignore the two incomplete evidence packages.
The authority arrangement helps identify a feasible response. The functional manager can assign another qualified reviewer or protect the existing reviewer's time. The product team can improve its local evidence handoff. It cannot appoint an unqualified reviewer, and it cannot waive independence. Starting more build work would not create additional review capacity. That could add pressure to the queue unless the team also addresses how completed-build work reaches review. The next action should address the observed constraint through the people who can change it.
There is immediate work as well as recurring work to address. Completing the two missing test records is a correction to the current condition. It helps those particular items become review-ready. Investigating why records were missing and changing a handoff practice to reduce recurrence can be corrective action. Protecting review capacity addresses the repeated diversion that contributes to the delay. Preventive action addresses a potential nonconformity before it occurs. An action is not automatically preventive merely because it could produce a benefit next week.
The distinction matters because a project can repeatedly fix individual items while leaving the same cause in place. It can also launch an ambitious improvement initiative while neglecting the work already affected. In this case, the owners prepare the missing evidence while the team and functional manager address the recurring conditions. These actions can proceed together within their authority. If an immediate safety or mandatory reporting issue were present, its required response would also proceed without waiting for a lengthy improvement investigation. The case describes a review-flow problem, so the response should stay proportionate to that evidence.
Extend this fictional situation with a small investigation before choosing the handoff change. Ask the author of each incomplete package to walk through how the test record was prepared and submitted. Assume, for this extension only, that one author used an obsolete template, while the other created the right record but stored it where the reviewer could not access it. Both packages appeared to have missing evidence, yet the mechanisms differ. Replacing the template addresses one mechanism. Correcting the approved access arrangement and linking the record addresses the other. Simply telling everyone to be more careful does not specify either repair.
Follow the evidence beyond the first plausible explanation. An obsolete template may persist because the current instructions point to two different locations. Restricted access may be intentional, in which case the right response is a permitted review route rather than opening the entire repository. Ask the relevant owners to confirm what happened, which constraint applies, and what change they can authorize. A cause diagram can organize possibilities, but drawing a box around an explanation does not verify it. Record unsupported explanations as hypotheses and choose a check capable of distinguishing them from alternatives.
The resulting action record should be specific enough for someone else to complete. In our extension, the documentation owner replaces the obsolete shortcut and verifies the current template with a user who normally follows that shortcut. The access owner confirms a permitted reviewer route and tests it using the reviewer's actual role. The process owner then watches whether those particular return reasons recur. A generic count of all returned items could hide progress on one cause while another worsens. Connecting each change to its observed failure mode makes the later review informative rather than ceremonial.
The proposed experiment lasts two weeks. It combines protected review slots with a clearer evidence handoff for the same item class. The functional manager agrees to the capacity arrangement, and the existing acceptance checks and independence requirement remain. The team records staffing, work mix, and other workflow changes. It compares the same measures using the same observation windows, then holds a decision review at the end. This experiment is permitted within the agreed boundaries. A proposal to alter mandatory controls would require a different authority route.
A good experiment also makes its learning objective explicit. Here, the team expects more reliable review capacity and more complete handoffs to reduce avoidable waiting and help ready items reach accepted completion. It should agree before evaluating the outcome how it will decide to continue, adapt, or stop. The case does not supply numerical decision thresholds, so we should not invent them and present them as approved. We can say that the decision must consider delivery flow, required quality, the capacity arrangement, and any adverse effects rather than judging throughput alone.
The two changes are tested together, which creates a practical limitation. If results improve, this comparison will not isolate precisely how much improvement came from protected time and how much came from clearer evidence. That does not make the test useless. The team is trying a feasible response to two observed conditions, not conducting a controlled laboratory study. It should record the combined change accurately, examine the supporting observations, and avoid claiming that one element has been proven to explain everything. Further refinement can follow if that distinction matters to a future decision.
Quality planning explains why the review remains in the flow. The acceptance conditions define the expected result and the evidence needed to evaluate it. The independent review contributes assurance required by the agreed arrangement. Product checks examine whether individual items meet specified requirements. Process improvement examines how work is performed and whether the method can produce the needed results more reliably. A successful process discussion does not make an unchecked item acceptable. Passing one item's tests does not establish that the whole delivery process is stable.
Cost-of-quality thinking adds another useful perspective. Clearer handoff guidance can be a prevention investment when it reduces the creation of incomplete or incorrect evidence packages. Reviewing and testing work to determine whether it meets requirements are appraisal activities. Repairing nonconforming work found before delivery is internal failure cost. Consequences of defects discovered after delivery can contribute to external failure cost. Classify an activity by what it does and when a defect is found. Automation that detects an existing defect is still appraisal; being automated does not make it prevention.
The case gives no dollar amounts for those activities, so there is no basis for announcing a financial saving. The team can examine time spent on returned work, repeated checks, review preparation, and operational consequences if reliable records are available. It should avoid counting the same effort twice or assuming every hour of waiting is paid rework. Waiting, reviewer effort, developer rework, and user delay describe different consequences. A useful economic comparison states its boundaries and preserves the required controls while comparing the total effects of a proposal.
We can add a separate economic exercise without pretending that the original case supplied dollar savings. Suppose a comparable handoff team processes ten packages per week. Before its change, four packages require additional preparation because required evidence is incomplete. In an illustrative follow-up, only one requires that extra preparation. Each affected package consumes one and a half hours of additional preparation, measured separately from the required independent review. The organization uses seventy-five dollars per staff-hour as a planning value. These are new teaching assumptions, not measured results from the six-week case we are following.
The original weekly preparation rework is four times one and a half hours, or six hours. The follow-up rework is one and a half hours, a reduction of four and a half hours. However, the improved handoff takes an additional half-hour of preparation each week across all ten packages. Deduct that extra effort. The net recurring capacity released is four hours per week, valued at three hundred dollars under the stated rate. Ignoring the new preparation effort would overstate the benefit. Counting independent review as removed work would contradict the requirement that it continues.
Assume the initial template correction and briefing take four staff-hours once, also valued at seventy-five dollars per hour. That is a three-hundred-dollar setup investment. Two weeks of the modeled recurring benefit equal six hundred dollars, leaving three hundred dollars after the setup effort. Under these unchanged assumptions, the modeled benefit covers the setup effort after one week. This is a simple effort-value comparison, not a discounted financial appraisal. It also assumes that the affected package counts and rework times remain comparable. If the new work is substantially easier, the apparent improvement needs further examination.
Released staff capacity is not automatically cash saved. Salaried employees may spend those four hours on other authorized work, making the benefit greater capacity or shorter delay rather than a smaller payroll payment. A cash-saving claim would need evidence that expenditure actually falls without moving the same expense elsewhere. Ask the manager how the time will be used and how that use will be observed. Avoid promising both a complete reduction in payroll cost and four extra hours of delivery from the same unchanged staffing arrangement. Those would be different benefit mechanisms requiring different evidence.
Now change one assumption. Suppose the improved handoff requires two and a half additional preparation hours per week instead of half an hour. The original reduction in rework remains four and a half hours, so net capacity released becomes two hours per week, valued at one hundred fifty dollars. It would take two weeks, rather than one, to cover the three-hundred-dollar setup effort. The idea may still be useful, but the decision now depends more on how long the benefit lasts. Sensitivity analysis helps identify which uncertain inputs deserve attention before a confident saving is advertised.
Now consider the experiment's actual follow-up results. In week five, eight items finish, seven remain in the review queue, median review wait is three working days, and one escaped defect is recorded. In week six, nine items finish, five remain queued, median review wait is two working days, and one escaped defect is recorded. Both defect observations use the same completed five-working-day follow-up windows described earlier. We are comparing accepted items under the same reported acceptance rule, rather than quietly counting build completion as finished work.
The movement is encouraging. From week four through week six, accepted completion rises from five to eight to nine. The end-of-week queue falls from nine to seven to five. Median review wait falls from five to three to two working days. Yet the queue has not disappeared. Five items remain waiting at the end of week six, compared with three at the end of week one. Median review wait returning to two days does not promise that every remaining item will start review within two days.
Take care when summarizing the two periods. The four baseline weeks contain twenty-six finished items in total. The two follow-up weeks contain seventeen. Comparing those totals alone would hide the different lengths of the periods. Average weekly throughput is six and a half in the baseline and eight and a half in the follow-up. Those averages help describe the observation, but they do not remove differences in work complexity or establish causation. The week-by-week sequence still matters because it shows the deterioration and subsequent change that an overall total would conceal.
The defect evidence deserves the same care. Baseline escaped-defect counts are one, one, one, and two; follow-up counts are one and one. These are counts, with different numbers of accepted items behind them. A defect count is not necessarily a count of distinct defective deliverables because one deliverable can have more than one defect. Severity and consequences would also matter to a quality decision. We should neither claim that quality is proven unchanged nor declare a precise reduction in defective-item percentage from data that do not establish that percentage.
The improvement owner does more than read the numbers. The owner interviews the reviewer and two developers, checks reasons that items were returned, and gathers feedback from the operational receiving team. There are no reported changes to the acceptance rule. However, week six includes somewhat simpler items. That is a concrete alternative explanation for some of the improved throughput or waiting time. It belongs in the recommendation. Reporting favorable measures while omitting a known change in work mix would make the conclusion sound stronger than the evidence warrants.
The appropriate conclusion is that this short follow-up supports a practical decision review. It does not experimentally prove that the changes caused every improvement. A reasonable recommendation is to consider retaining the protected review arrangement and clearer handoff within the approved boundaries, continue comparing consistent measures, and examine whether the benefits persist with a representative work mix. That recommendation should acknowledge the remaining queue and any effort or disruption created elsewhere. The process owner and team can then decide whether the observed value justifies continued use or a further adjustment.
Consider a hypothetical variation, separate from the case results. Suppose throughput rises, but the team discovers that an independent check was skipped. The higher completion number would not demonstrate acceptable improvement under this case's rules. The affected work would need the appropriate review and response, and the reported result would need correction. In another hypothetical variation, suppose the queue falls only because incoming work stopped. The team would need that arrival information before crediting the review change. These variations illustrate why measurement definitions, control integrity, and context travel with the numbers.
A queue also needs its oldest unfinished work made visible. As a separate invented snapshot, suppose five items have been waiting for one, two, two, three, and twelve working days. Their median age is two days, yet the oldest item has waited twelve. This age distribution is not the completed-item waiting measure reported in the original case. One describes work still waiting now; the other describes waits associated with items that reached review. A team can improve the completed-item median by finishing newer, easier work while an older difficult item remains stranded.
Ask what keeps the twelve-day item waiting and who can resolve that condition. It may lack a permitted specialist, depend on an external decision, or have an incomplete evidence package. The record should distinguish those possibilities instead of treating age as proof that a particular reviewer is at fault. Prioritizing old work also does not mean ignoring a newly arrived safety-critical item. Apply the agreed service and risk rules, and make exceptions visible. The purpose of an aging review is to expose neglected work and obtain a feasible next step, not to create a universal first-in, first-out rule.
Check the queue's arithmetic as well as its appearance. In another small illustration, a queue starts with nine items, receives seven, and has eleven leave that state during the observation period. With no other additions, removals, or reclassifications, it ends with five. If the displayed total is four, reconcile the missing item before claiming extra improvement. Leaving a review queue is not necessarily accepted completion; an item could enter active review or return for correction. Preserve those transition definitions so moving cards between columns cannot manufacture a delivery result that never occurred.
Improvement becomes durable when the supported learning changes how future work is done. If the change is retained, the process owner can approve updated handoff guidance, state where it applies, and schedule a review of continuing use. The guidance might explain what a complete evidence package contains and how ready work reaches a protected review slot. It should reflect the actual approved arrangement. A checklist cannot substitute for a capacity agreement, and a capacity agreement cannot tell an author which required test record belongs in a package.
Give the maintained guidance an owner, an effective version, an approval record, and a clear place where people find the current instructions. Record the context of the experiment and the limits on applying it elsewhere. Historical versions can remain available where they explain earlier decisions, while the normal working location points to the current version. Ask someone who did not write the guidance to use it on a representative handoff. If that person still needs the author's private explanation, the asset needs further work before it can support repeatable delivery.
Transition creates the final test. The project will eventually transfer the digital service and its support arrangements to operations. The receiving organization needs continuing review capacity, current handoff guidance, and responsibility for future improvement. It also needs to know who owns the service outcomes that motivated the project. The flow measures show how project work moved toward acceptance. They do not, by themselves, prove reliable intake or less avoidable waiting for users. Those operating outcomes require their own agreed evidence and continuing owner after the project team leaves.
For your study practice, reconstruct the decision without memorizing a slogan. Define finished work, identify the growing queue, separate missing evidence from unavailable qualified capacity, and locate the relevant authority. Describe the authorized experiment and compare its results on the same basis. Then name the simpler week-six work as a limitation and explain what the process owner should review. Each part supports the next. The answer is stronger because it connects observed conditions to an executable response and a maintained process, rather than treating a retrospective note as proof that learning happened.
Continuous improvement earns its place through that complete cycle of observation, action, review, and continued ownership. In this case, the team can respond to a real bottleneck while preserving independence, reduce avoidable handoff friction, and make an appropriately qualified recommendation from a short comparison. The next lab turns to organizational change and adoption. We will move from improving the team's delivery process to examining whether the people receiving a changed service can use it, support it, and carry its intended benefits into everyday work.
For more certification study resources, books, and podcasts, visit Bare Metal Cyber dot com. Thank you for listening to Certified, the P M P prepcast.