WHY ARE PROJECTS ALWAYS LATE? (what can the Project Manager DO about that?) Craig Henderson, MBA, PMP ARVEST Bank Operations
Introduction PM Basics FIO GID KISS (Figure it out) (Get it done) (Keep it simple, sweetie) Predictable Project Delivery - Quality product - On time - On budget
Planning FIO Detailed, precise SCOPE Create WBS structure Estimate activity durations Sequence activities Network diagram Develop schedule With all this wonderful work, WHAT COULD POSSIBLY GO WRONG??
Planning Estimate Errors? Scope Creep? Execution? Risks?
Estimate Errors? Factors influencing estimate quality Planning horizon Immediate events more accurate than distant events Project duration Shorter durations more accurate than long People Resource skill levels Team experience Turnover Productive time (5-6 hours/day?) Project structure & organization Estimate padding (or Understating?!) Organization culture
Estimate Errors? Estimate Errors? Tradeoffs Nothing is free (but everything else Costs!) Estimates are not free Better estimates are more expensive Must balance accuracy/cost tradeoff Don t underestimate the estimate
Task Data task pred 1 pred 2 Estimate A 3 B A 2 C A 5 D A 4 E B 3 F C 2 G D 8 H E F 4 I H 4 J G 4 K I J 7 What is your project duration estimate?
(network diagram reminder) AON (Arrow on Node) ES ID EF SLACK LS Dur LF ES ID EF ES ID EF SLACK SLACK LS Dur LF LS Dur LF ES ID EF SLACK LS Dur LF
Network Diagram 5 E 8 3 B 5 3 6 2 8 3 8 3 11 10 H 14 1 11 4 15 14 I 18 1 15 4 19 0 A 3 8 F 10 0 0 3 3 3 C 8 1 4 5 9 1 9 2 11 19 K 26 0 19 7 26 3 D 7 0 3 4 7 7 G 15 0 7 8 15 15 J 19 0 15 4 19 What is the Critical Path?
Critical Path 3 B 5 3 6 2 8 5 E 8 3 8 3 11 10 H 14 1 11 4 15 14 I 18 1 15 4 19 0 A 3 0 0 3 3 3 C 8 1 4 5 9 8 F 10 1 9 2 11 19 K 26 0 19 7 26 3 D 7 0 3 4 7 7 G 15 0 7 8 15 15 J 19 0 15 4 19
Task Data task pred 1 pred 2 Estimate A 3 B A 2 C A 5 D A 4 E B 3 F C 2 G D 8 H E F 4 I H 4 J G 4 K I J 7 CP= Tot CP Est 26 A-D-G-J-K What are our odds of finishing on time, P(26)? 100%? 50% (50/50)? (Something else?)
Not So Fast! Task Duration One point? Three point? Distribution? 95% Payback, BEST IN TOWN! (Fitzgerald s Casino, Reno, NV) Feeling lucky, Sucker?!
PERT Program Evaluation Review Technique Assumes activity duration is a range statistically following a beta distribution. 3 time estimates for each activity: Expected Optimistic Pessimistic Weighted average represents activity duration distribution. Weighted average and variance for each activity allows computed probability for various project durations.
Activity and Project Frequency Distributions
Activity Time Calculations The weighted average activity time is computed by the following formula:
Activity Time Calculations (cont d) Activity time estimate variability approximated by: Standard deviation for activity: Standard deviation for project: Note standard deviation of activity is squared in this equation; also called variance. This sum includes only activities on the critical path(s) or path being reviewed.
Task Data task pred 1 pred 2 a m b A 2 3 4 B A 1 2 3 C A 4 5 12 D A 3 4 11 E B 1 3 5 F C 1 2 3 G D 1 8 9 H E F 2 4 6 I H 2 4 12 J G 3 4 5 K I J 5 7 8 CP= A-D-G-J-K
Task Times task a Est (m) b T(e) sigma sigma^2 A 2 3 4 3 0.333333 0.111111 B 1 2 3 2 0.333333 0.111111 C 4 5 12 6 1.333333 1.777778 D 3 4 11 5 1.333333 1.777778 E 1 3 5 3 0.666667 0.444444 F 1 2 3 2 0.333333 0.111111 G 1 8 9 7 1.333333 1.777778 H 2 4 6 4 0.666667 0.444444 I 2 4 12 5 1.666667 2.777778 J 3 4 5 4 0.333333 0.111111 K 5 7 8 6.833333 0.5 0.25 Totals: 26 25.83333 4.027778 CP= A-D-G-J-K CP sigma = 2.006932 66% of the time, this project will complete in the range of??
Probability of Completing the Project @ X Compute the Z value (Z = number of standard deviations from the mean) Then find probability of Z
Z Values and Probabilities
Possible Project Duration Example Probability project is completed before scheduled time (T S ) of 67 days Probability project is completed by the 60 th day (T S )
Probability of finish by Est? Z = ( Mean value) / = (26 25.83) / 2.01 =.083 p(26) = 53% StndDev
Possible Project Duration Probability project is completed before scheduled time (T S ) of 26 units Total Z = ( Mean value) / = (26 25.83) / 2.01 =.083 p(26) = 53% StndDev 25.83 26 53%
Monte Carlo Simulation Randomize task times Random normal number Adjusted for µ & σ of each task s distribution Add resulting task times per network diagram Do this many times! Calculate average (expected value) and σ
Monte Carlo Simulation in Excel Generate random results for each task Analysis Tool Pack, Random Number Generator Following CP, add times Remember IF statement to check CP length! Calculate average project time/stnd dev Compare to previously computed result
Monte Carlo Results T e = 27.57 Stnd Deviation = 1.83 (Remember, Computed T e = 25.83 Stnd Dev = 2.0) % change (mean)= 6.36% % change (sigma)= -8.72% How can this be?
Monte Carlo CP = A-D-G-J-K BUT Monte Carlo runs showed CP finish of A-C-F-H-I-K = 61.3% Observation CP shifted from original path about 2/3 of the time! The CP shift prevented us from gaining full advantage when the original randomized CP was very early. What is the term for a network like this? CP Results # % A-B-E-H-I-K= 15 1.5% A-C-F-H-I-K= 613 61.3% A-D-G-J-K= 372 37.2%
Critical Path, and others 3 B 5 3 6 2 8 5 E 8 3 8 3 11 10 H 14 1 11 4 15 14 I 18 1 15 4 19 0 A 3 0 0 3 3 3 C 8 1 4 5 9 8 F 10 1 9 2 11 19 K 26 0 19 7 26 3 D 7 0 3 4 7 7 G 15 0 7 8 15 15 J 19 0 15 4 19
Monte Carlo Result - Beta Distribution? Total Count of beta CP Results # % A-B-E-H-I-K= 15 1.5% A-C-F-H-I-K= 613 61.3% A-D-G-J-K= 372 37.2% beta
Task Data task pred 1 pred 2 a m b A 2 3 4 B A 1 2 3 C A 4 5 12 D A 3 4 11 E B 1 3 5 F C 1 2 3 G D 1 8 9 H E F 2 4 6 I H 2 4 12 J G 3 4 5 K I J 5 7 8 CP= A-D-G-J-K
Management s Project Target Total Count of Random Z Random Z Est CP = 26, T e (computed) = 25.83, T e (simulation) = 27.57
Confidence and Completion 95% confidence project will complete by time X 95% Payback, BEST IN TOWN! (Fitzgerald s Casino, Reno, NV) = Z *( StndDev) + Mean = 1.64*1.83+ 27.57 = 30.57 Feeling lucky, Sucker?!
Possible Project Duration 95% confidence Project Duration = Z *( StndDev) + Mean = 1.64*1.83+ 27.57 = 30.57 Total 27.57 95% 30.57
Probability of finish by Est? P(26)? Z = ( Mean value) / = (26 27.57) /1.83 =.858 p(26) 20% StndDev Feeling lucky, Sucker?!
Probability of finish by Est? Total Z = ( Mean value) / = (26 27.57) /1.83 =.858 p(26) 20% StndDev 27.57 26 20%
Project Result Summary CP T(e)= 25.83 CP sigma = 2.00 Project 95% con= 29.12 P(26) = 53.32% rlo T(e)= 27.57 sigma = 1.83 Project 95% con= 30.58 P(26) = 20%
Remember Activity definition and precedence Network diagram Task estimates, and ranges/distributions Calculate Expected Durations on CP Consider Monte Carlo simulation Finally: Calculate Management Duration based on desired confidence/predictability Add this to your Risk Plan and Risk Budget
Why are projects always late? Project Managers frequently fail to understand and account for Actual result distributions vary from (point) estimates In their Risk Plans and Budgets Feeling lucky, Sucker?!
Questions! Feeling lucky, Sucker?!