Quality tools

0:00Sara: Welcome to Project Management Exam Prep.

0:03This episode is about quality tools: finding the causes of a problem,

0:07ranking them, and watching a process over time.

0:10Leo: Every project management exam can ask for them.

0:14You will learn four tools, and how to choose between them.

0:17Quality planning, assurance and the cost of quality have their own episode.

0:22We finish with a drill, so keep a calculator ready.

0:25Sara: Let us start with the three lenses. What does IPMA® say?

0:29Leo: IPMA®, the International Project Management Association, sets out its standard in

0:35the Individual Competence Baseline, the ICB4.

0:39Its element on quality asks you to find defects with established tools,

0:43analyse their causes, and explain a root-cause analysis.

0:48Its element on resourcefulness adds analytic techniques for causes and trends.

0:53Sara: And PMI®?

0:54Leo: PMI®, the Project Management Institute, publishes the PMBOK® Guide,

1:00its guide to the project management body of knowledge.

1:04The eighth edition lists the cause-and-effect diagram, the five whys and control

1:09charts among its tools. The outline of PMI®'s Project Management Professional exam,

1:15the PMP®, asks you to plan quality processes and tools, and to improve continuously.

1:22Sara: And APM?

1:24Leo: APM, the Association for Project Management, publishes the APM Body of Knowledge.

1:31There, quality control documents and remedies defects, and continuous improvement

1:36starts from performance data. The syllabus of APM's Project Management Qualification,

1:42the PMQ, asks how quality control techniques show whether success criteria are met.

1:48Sara: Where do the words differ?

1:50Leo: In two places. First, the search for causes.

1:54The ICB4 and the PMBOK® Guide say root cause analysis.

1:59APM's chapter on data files it under diagnostic analytics: why did it happen?

2:05Sara: Second?

2:06Leo: The limit a requirement sets. PMI® says specification limit, APM acceptance criterion,

2:15and the ICB4 target value for a quality indicator.

2:18None of them is a control limit.

2:22Sara: What is the core, whichever exam you take?

2:25Leo: Choose the tool by the question. One: why does it happen?

2:30A cause-and-effect diagram goes wide, and the five whys go deep.

2:35Two: which problem first? A Pareto chart ranks the causes.

2:41Three: is the process stable, or has something changed?

2:44A control chart tells noise from a signal.

2:48Always put data before opinion: count before you fix, and after.

2:54Sara: And in agile work?

2:56Leo: A team can draw a quick fishbone in a retrospective, or count why items came back

3:01from testing. APM names retrospectives among the tools of continuous improvement.

3:07The PMBOK® Guide also puts cost and schedule variances on control charts.

3:12Sara: Where do I start?

3:12Leo: With the effect, as a fact with a number.

3:16Take a new online booking service: in its first month, two hundred bookings fail.

3:22Write it in a box on the right, with a spine to it.

3:25Each rib holds a group of possible causes, filled in by the team.

3:29Kaoru Ishikawa, a Japanese quality pioneer, made this fishbone diagram popular.

3:35Sara: And the five whys?

3:37Leo: They go deep, down one rib, as Taiichi Ohno, the engineer behind Toyota's production

3:43system, described. Why do bookings fail?

3:47The payment step times out. Why? It waits for an address check.

3:51Why is that slow? It calls an outside service for every booking.

3:56And why was that not caught? Nobody tested at peak load.

4:01So speed up the check, and add a load test to every release: one corrective action,

4:05and one preventive.

4:06Sara: Which rib matters most?

4:10Leo: Count. A fishbone shows what could cause the effect, not how often.

4:15Joseph Juran applied the Pareto idea to quality in nineteen fifty-one:

4:20a vital few causes produce most of the losses.

4:24He named it after the economist Vilfredo Pareto, and later called the name his mistake.

4:30One bar per cause, the largest first. Timeouts: ninety.

4:35Rejected addresses: fifty. Declined cards: thirty.

4:39Three small causes: thirty.

4:42Sara: And the line?

4:43Leo: The running total, as a share. Ninety of two hundred is forty-five percent.

4:48Add fifty: seventy. Add thirty: eighty-five.

4:53That is three causes of six, so eighty-twenty, the idea that a fifth of the causes

4:58give four fifths of the effect, is only a rule of thumb.

5:01Fix where the line bends.

5:04Sara: The fix is live. How do I know it holds?

5:07Leo: With a control chart, which Walter Shewhart set out in nineteen thirty-one.

5:12Plot one measure over time, with its mean.

5:15By a common convention, the control limits sit three standard deviations either

5:20side of the mean of a stable period. Say failed bookings now average ten a day,

5:26and the standard deviation is two. Three times two is six.

5:30So the limits are sixteen and four.

5:33Sara: What does it tell me?

5:34Leo: Which variation to leave alone. Points between the limits are common cause variation,

5:40the normal noise. Chasing them adds noise.

5:44A point outside signals a special cause: something changed, so find it.

5:50On day eleven, eighteen bookings fail: the payment partner had changed a setting

5:55that morning.

5:57Sara: And inside the limits?

5:59Leo: Watch for runs. A common rule of thumb, the rule of seven: seven consecutive points,

6:05all above the mean or all below it. From day fourteen, seven days sit above ten.

6:12All are inside the limits, but the process has shifted.

6:16Sara: Why seven?

6:18Leo: In a stable process, each point is a coin toss, above or below the mean.

6:23Seven heads in a row is one half, seven times over.

6:27That is one chance in a hundred and twenty-eight.

6:31Sara: So inside the limits is good enough?

6:34Leo: Not always. Control limits come from the data; specification limits,

6:39from the requirements. Here the service level allows at most thirteen failures a

6:45day, below the upper limit of sixteen. So a stable process can still miss it.

6:51Then change the process.

6:53Sara: How does this change between Mid-level and Senior?

6:57Leo: First, the two levels. They are this series' own labels.

7:00Mid-level means leading moderately complex projects, the level of IPMA® Level C and

7:06PMI®'s PMP® exam. Senior means leading complex projects and people,

7:12the level of IPMA® Level B and of APM's chartered status, Chartered Project Professional.

7:19Take Parking Permits Online, a city project that moves resident parking permits

7:23online in eight months. After go-live, the project manager, Anna,

7:28has the service desk log each problem with one cause.

7:32She ranks the causes, runs the five whys with the supplier, and watches one control

7:37chart through January, when most residents renew.

7:41Sara: And at Senior level?

7:42Leo: You design how many parties measure and react.

7:47Take One City Account, a thirty-month city project that brings fourteen online services

7:53under one login, led by Anna some years later.

7:57Anna agrees one list of defect causes for every service and both main suppliers,

8:04so their charts compare. Each key measure has agreed signals,

8:09an owner and a time to react. The steering board sees trends,

8:14not single bad days.

8:17Sara: Now the drill. A city moved resident parking permits online four weeks ago,

8:23ahead of January, when most residents renew.

8:26Each day, the team counts residents who give up an online renewal before paying.

8:30Over twenty working days, the mean is fifteen a day, and the standard deviation

8:35is two. The steering group's target is at most twelve.

8:40Yesterday, twenty gave up, and the sponsor wants the supplier to explain it by tomorrow.

8:46Is yesterday a signal? What do you do?

8:49Leo: Pause and write your answer. You have forty-five seconds.

9:37Sara: First, the chart.

9:39Leo: Three times the standard deviation of two is six.

9:43So the limits are nine and twenty-one. Twenty is inside: common cause.

9:50Unless there is a run of seven, chasing yesterday wastes the supplier's time.

9:54But the target of twelve is below the mean of fifteen.

9:57The process is stable, but misses the target on most days.

10:02Sara: Why most days?

10:04Leo: In a stable process, about half the days fall above the mean.

10:08Twelve is below it, so even more days miss.

10:11Only a change to the process moves the mean.

10:14So use a Pareto chart of where residents give up, the five whys on the tallest bar,

10:19and a new control chart for the new mean.

10:22Sara: What lifts that to Senior?

10:23Leo: January brings far more renewals, so chart the share who give up,

10:29not the count. Agree in advance with the steering group which signals start action,

10:35who acts, and how fast. Put the target in the supplier's service level.

10:40Weigh a fix before January against extra staff through it.

10:45And treat common cause variation as the system's problem, not a person's fault.

10:50Sara: How do the exams ask this?

10:53Leo: IPMA®'s exams ask for open answers. IPMA®'s certification also includes an interview,

10:59where assessors ask about your own projects: note one root cause you found,

11:05and your evidence. The PMP® exam uses scenarios, and questions built on a case study

11:10or a chart. APM's PMQ asks for short written answers, such as how quality control

11:16shows that success criteria are met.

11:19Sara: Let us recap.

11:20Leo: One. Choose the tool by the question.

11:23Sara: Two. The fishbone goes wide; the five whys go deep.

11:27Leo: Three. A Pareto chart ranks the causes; the cumulative line shows the vital few.

11:35Sara: Four. A point outside the control limits, or seven consecutive points above or below

11:41the mean, signals a special cause.

11:43Leo: Five. Control limits are not specification limits.

11:48Stable but off target? Change the process.

11:51Sara: The three model sheets and the study handout are free in the description.

11:55Next time: agile planning and metrics.