1. Formulating a Problem for Systems Analysis

2. Steps to uncover structures Formulate the problem Identify key variables in the situation (key structural pieces and their relationships)

3. Steps for uncovering structures (cont.) Graphing the behavior of those variables over time

4. Systems thinking is an experimental process Expect trial and error. Engage an iterative process. Create a problem statement. Offer hypotheses about what’s happening.....

5. Systems thinking is an experimental process (cont.) Track and revise reasoning behind explanations. Test possible solutions. Reformulate the problem based on new understandings.

6. Formulating a problem Determine whether the problem is a “systemic problem” The problem is chronic and recurring. The problem has been around long enough to have a history.

7. Formulating a problem (cont.) You or someone else has tried to solve it, but your attempts failed, or stopped working after awhile. You haven’t been able to identify an obvious reason for the pattern of behavior over time. The pattern of the behavior over time is classic (see next slide).

8. Example of a problem that shows typical systemic behavior We’ve been having trouble getting our refrigerators assembled fast enough to fill customer orders. We reconfigured the flow of materials on the manufacturing floor to improve the assembly process. However, after we made this change, the assembly process took more time than before. Somehow the change made everything worse.

9. Guidelines for formulating a problem Make the description concise. Be clear. Formulate 2 or 3 versions that differ. Expect that this stage will take awhile.

10. Possible components of problem statements The behavior (customer service problems) A description of the behavior over time (the problems have increased)

11. Possible components of problem statements (cont.) A measurement of how the behavior has changed over time (the problem has increased 50%) The time frame of the behavior (this year)

12. Final problem statement Customer service problems have increased 50% this year, over last.

13. Identifying variables Variables of the problem are components whose value can go up or down over time. Make a preliminary list and then count on adding or deleting some.

14. Guidelines for identifying variables List all the variables that should reasonably be included. Get input from multiple perspectives.

15. Guidelines for identifying variables (cont.) Narrow the list to the most important. See which are most firmly linked to the central problem.

16. Guidelines for naming variables USE--- New product designs Revenues Experience of designers NOT--- Designing new products Selling products Being profitable Use nouns or noun phrases, not verbs.

17. Guidelines for naming variables The number of new products in design The amount of revenue The size of profit margin A well-named variable will fit into phrases like: the level of; the amount of; or the number of.

18. Guidelines for naming variables “Good design” rather than “Bad design” “Trust” rather than “Lack of trust” (If the variable increases, improves, etc., you can then address the change without creating a double negative.) Use a neutral or positive term when possible to name a variable.

19. One more thing… Variables can be concrete things such as widgets... Or intangibles such as alignment with company vision.

20. Drawing Behavior Over Time Graphs (BOTs) Use 3 steps: 1. Select a time horizon 2. Sketch the graph 3. Build theories about how the variables on the graph are related

21. Selecting a time horizon The time span you look at affects your sense of the problem and its solution Pick the variable with the longest time- cycle, e.g. a sales or design cycle Extend the time horizon to examine at least 3 of those

22. Selecting a time horizon (cont.) Try to work with a minimum of 2 years Experiment with 5 or more years If you think you need a much shorter or longer time horizon, check with colleagues to reason why

23. Selecting a time horizon Sketch a timeline like this: Now is the present moment when you are analyzing the problem. Earlier is the point when you will begin tracing the behavior of these variables. Earliest is a point even earlier in time, when something that happened may have started the problem. EarlierEarliestNowFuture

24. Selecting a time horizon (cont.) Think about each related variable and its behavior over this timeline Imagine how each will act if no intervention is made Use this to help you visualize the graph

25. Sketching the graph Graph the variables together so you can see their interrelationships over time Look for how variations in behavior might be dynamically related On the horizontal axis, write the time units you used, e.g. years, dates, etc.

26. Sketching the graph (cont.) If there was a single, high-impact event that happened during this time, draw and label a vertical line on the graph and show when it occurred.

27. Building testable hypotheses Now, hypothesize about what you see. Be ready to add variables, draw more graphs and so on. Remember - BOT graphs are just a visual thinking tool - not a masterpiece of hard data!