itsmf Finland Conference 2015 It Feels Good! Phil Green In God We Trust Everyone Else Bring Data

Size: px

Start display at page:

Download "itsmf Finland Conference 2015 It Feels Good! Phil Green In God We Trust Everyone Else Bring Data"

Philip Ellis
5 years ago
Views:

1 itsmf Finland Conference 2015 It Feels Good! Phil Green In God We Trust Everyone Else Bring Data

2 When people are pressured to meet a target value, how might they proceed? 1 They can work to improve the system They can distort the system They can distort the data 1 Brian Joiner, Fourth Generation Management, McGraw-Hill

Things aren t always what they seem! The table tops do not look alike but they are!

February deficit 1.6 billion smaller than previous February things are better!

1 - Source: Roger N Shepard, Mind Sights: Original Visual Illusions, Ambiguities and Other

3 Things aren t always what they seem! The table tops do not look alike but they are! 1 Good or bad? February deficit 1.4 billion larger than January things are worse! February deficit 1.6 billion smaller than previous February things are better! 2 These are examples of a small number of data points or observations without proper context. 1 - Source: Roger N Shepard, Mind Sights: Original Visual Illusions, Ambiguities and Other Anomalies, WH Freeman and Company 2 - Derived from Understanding Variation: The Key to Managing Chaos, Donald J. Wheeler, SPC Press Inc.

4 A Case in Point INVENTORY IN DEPARTMENT 50 Derived from Understanding Variation: The Key to Managing Chaos, Donald J. Wheeler, SPC Press Inc.

5 Inventory in has fallen to the lowest level recorded in three years! The Manager rewards everyone with a celebration!

6 But then inventory rises again The Manager regrets giving the award time to have a serious talk with the team!

7 Inventory levels fall again The Manager thinks tough talk gets results!

8 But then inventory rises to the highest level recorded in two years! Manager throws a tantrum, demands improvement plans. The team keep a low profile, hoping things get better, hiding inventory in dark corners, under their desks, etc.

9 And once again Inventory falls Once again, the manager thinks tough talk works!

10 But then inventory rises yet again And around and around we go. Lather, rinse, repeat!

11 So what went wrong? Team needs talking to! Here we go again! Tantrum! Award Given Tough talking works! They re finally getting the message! We ve been reacting to data points, comparing numbers to specifications the Voice of the Customer (VoC) which hasn t worked. To understand, we need to listen to the Voice of the Process (VoP)

12 VoP The Control Chart Approach UCL CL LCL So what is wrong with this process? Statistically, nothing! The process is in control not one of the data points is a signal to be concerned about. The process shows only common cause variation A change in process requires a structured improvement such as Plan-Do-Check- Act (PDCA), not reacting to normal data points (known as tampering! )

13 Anything noticeable here? A signal an assignable cause

14 And here? A set of consecutive points below the centre line. A change in the process has occurred

15 Control Limits can be changed Centre line has moved Process variation has been reduced

16 What have we learned? Without proper context, data is meaningless! Comparing data points to specifications (VoC) and reacting is merely tampering and won t drive sustainable improvement The VoC defines what you want from a system, whereas the VoP defines what you will actually get A control chart will shows us the VoP The VoP will filter out the noise (common cause variation) so we can clearly see the signals (assignable cause variation)

17 Six Sigma Overview A SUMMARY OF THE THEORY

18 Introduction to Six Sigma Pioneered by Motorola in the 1980 s, providing $16 Billion in savings Widespread adoption since, e.g., GE, Ford, Honeywell, GSK, Nokia Proven to yield dramatic improvements in business results Lean (Six) Sigma drives process improvement by reducing waste Six Sigma drives improvement by reducing variation

19 Variation why does it bother us? Some things varying can be ok, but not others: Quality of food or service at your favourite restaurant? Unreliable buses or trains? Call wait time telephoning your bank? Produce freshness at the food store? Cycle time from order to delivery of items needed to manufacture goods?

Sigma as a Quality Measure DPMO = defects per million opportunities for error A Case in Point: Concert tickets printed 5 opportunities for error (venue, date, act, time, seat) 1000 tickets issued, 50

20 Sigma as a Quality Measure DPMO = defects per million opportunities for error A Case in Point: Concert tickets printed 5 opportunities for error (venue, date, act, time, seat) 1000 tickets issued, 50 defects DPO = 50/(5 x 1000) = 0.01 Yield = (1-0.01) x 100 = 99% DPMO = 0.01 x 10 6 = 10,000 (3.8σ achieved 1 ) Rule of thumb: You need 4σ or higher to satisfy most customers 2 1 Research has shown that the centre of a processes can naturally shift over time by up to ±1.5σ 2 Joe Parito, isixsigma

Control Chart Basics Centre line is the arithmetic mean of the data points Limits are 3σ (3 standard deviations 1 ) either side of CL UCL/LCL are not specification limits The Empirical Rule About

21 Control Chart Basics Centre line is the arithmetic mean of the data points Limits are 3σ (3 standard deviations 1 ) either side of CL UCL/LCL are not specification limits The Empirical Rule About 60-75% of the data points will be ±1σ from the CL About 90-98% of the data points will be ±2σ from the CL About % of the data points will be ±3σ from the CL Control limit calculations: Assignable Cause Area Upper Control Limit (UCL) Centre Line (CL) Lower Control Limit (LCL) Assignable Cause Area Voice of the Customer 1 Standard deviation = mean of (variation of each point from CL) 2 Voice of the Process

22 Control Chart: Case In Point Example UCL = 88.3 CL = 60.3 LCL =

23 Looking for Assignable Causes One or more points outside the control limits

24 Looking for Assignable Causes Eight or more consecutive points the same side of the Centre Line

25 Looking for Assignable Causes Sixteen points alternating up and down (rare) Other repeating patterns should also be investigated A rule of thumb: a pattern repeating 8 times

26 Learning Models DMAIC & DMADV DMADV is also known as Design for Six Sigma This presentation will discuss the DMAIC model.

27 The DMAIC Improvement Cycle Define articulate the problem quantifiably, the underlying process and the goal for improvement. Measure collect data to establish current baseline, and to identify & monitor the gap between current and required performance. Analyze identify and prioritise potential root causes for elimination to close the gap. Use statistical tools to guide the analysis. Improve identify and implement (creative) solutions; use statistical methods to validate the improvement. Control identify and implement sustainment strategies that ensure the improvement is embedded and sustainable.

Tools & Techniques Study Results: Control Charts / SPC Pareto FMEA Standardise: Control Chart / SPC FMEA Reporting Systems Implement: Proto type / Pilot studies Gantt Chart Matrix Test Theories:

28 Tools & Techniques Study Results: Control Charts / SPC Pareto FMEA Standardise: Control Chart / SPC FMEA Reporting Systems Implement: Proto type / Pilot studies Gantt Chart Matrix Test Theories: Ishikawa Diagram Process Maps Control Charts / SPC Brainstorming Matrix Design of Experiments Simulations Analyse Select Priorities: Affinity Diagrams System / Process Map Benchmarking Project Charter Pareto Learn About Process: Run Chart Process Map (e.g., flow chart) Histogram SIPOC VoC tools (surveys, focus groups, etc.) Investigate Sources of Variation: Run Chart Ishikawa Diagram Pareto Scatter

29 Incident Resolution Improvements Application Support Service SERVICE OPERATIONS CASE STUDY

DMAIC Define Stage Select Priorities Shared Application Support Service Several hundred applications, supporting 90K+ business users across 100+ countries Target agreed with business

30 DMAIC Define Stage Select Priorities Shared Application Support Service Several hundred applications, supporting 90K+ business users across 100+ countries Target agreed with business to drive incident resolution 90 th percentile down to 40 hours or below. Learn About Process Baseline compliance to incident resolution target 79% Baseline 90 th percentile 263 hours

31 SIPOC

32 DMAIC Measure Stage Investigate Sources of Variation Techniques selected Brainstorming Cause & Effect (Ishikawa) diagram Pareto charts

33 Cause & Effect (Ishikawa) Diagram

34 Pareto Chart

35 DMAIC Analyse Stage Form & Test Theories Not following good ITSM practice Closing incidents on the basis that a corresponding Problem record was raised Incorrect hold times calculations Clock was being stopped Changes logged as Incidents! Categorisations in system not clear

36 Root Causes Selected to Address

37 DMAIC Improve Stage Implement Re-training in good ITSM practice Change on hold time calculations System usability adjustments Study Results Incident resolution control chart 90 th percentile control chart Reactivation rates control chart (Balancing measure)

38 Incident Resolution Control Chart Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07 UCL=97.26 UCL= CEN=90.8 LCL=84.34 CEN= LCL= Percent Incidents Resolved in SLA Signal: 8 consecutive points above CL Moving R Chart UCL= UCL= CEN=4.0 CEN= LCL=0.0 LCL=0.0 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

39 Control Chart Limits Re-drawn Percent Incidents Resolved in SLA Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07 UCL= UCL= CEN= CEN= LCL= UCL= CEN= LCL= LCL= Moving R Chart UCL= CEN=4.0 UCL= CEN=3.5 UCL= CEN= LCL=0.0 LCL=0.0 LCL=0.0 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

40 90 th Percentile Control Chart

41 Reactivation Rates Control Chart

DMAIC Control Stage Improvements 90 th percentile improved 82% from 263 to 45 hrs Incident Resolution SLT compliance raised from 79% to 92% Reactivation Rates reduced from

42 DMAIC Control Stage Improvements 90 th percentile improved 82% from 263 to 45 hrs Incident Resolution SLT compliance raised from 79% to 92% Reactivation Rates reduced from 8% to 3% Sustainment Actions New practices formalised within procedures, knowledge base, training materials, etc. Training and communication ITSM tool changes made permanent

43 Additional Resources LinkedIn Groups Lean Six Sigma (and sub-groups) Lean Six Sigma Worldwide Useful Websites Computing Resources SPC XL Software (Excel plug-in, Windows only) QI Macros (SPC & Statistical software for Mac) Excel (calculates mean and standard deviation) Graphical Calculators Training/Certifications Books Donald J Wheeler, Understanding Variation, The Key to Managing Chaos, SPC Press Thomas Pyzdek, The Six Sigma Handbook, McGraw-Hill George/Rowlands/Price/Maxey, The Lean Six Sigma Pocket Toolbook, McGraw-Hill Quentin Brook, Lean Six Sigma and Minitab: The Complete Toolbox Guide for All Lean Six Sigma Practitioners, OPEX Resources

44 Questions? Phil Green Director, G3 Service Solutions Limited LinkedIn:

45 Thank You!

Visit us at:

Visit us at: White Paper Integrating Six Sigma and Software Testing Process for Removal of Wastage & Optimizing Resource Utilization 24 October 2013 With resources working for extended hours and in a pressurized environment,