From Peter Senge's levels of explanation (The Fifth Discipline, 1990), drawn as an iceberg by Michael Goodman (Systems Thinking: What, Why, When, Where, and How?, 1997).
Reads a problem at four levels: the visible event, the pattern it belongs to, the structures producing the pattern, and the beliefs that keep those structures in place.
Eight observations about missed quarter-end shipments, sorted by how deep they sit
Select an observation, then the level it belongs to: the event at the surface, the pattern it repeats, the structure that produces the pattern, or the belief that keeps the structure in place. Each level says what acting there would change. The first is placed as an example. Show the sort to compare yours. Illustrative.
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
- Not placedPlacedEventPatternStructureBelief
Depth
What the team noticed
Your sort matched 0 of 8.Your sort matched 1 of 8.Your sort matched 2 of 8.Your sort matched 3 of 8.Your sort matched 4 of 8.Your sort matched 5 of 8.Your sort matched 6 of 8.Your sort matched 7 of 8.Your sort matched 8 of 8.
- Event
- What happened. Acting here is reacting: expedite the shipment, pay the overtime. It fixes this quarter's miss and nothing after it.
- Pattern
- What keeps happening. Acting here is anticipating: staff the plant for the last two weeks of every quarter. The spike still comes; it costs less.
- Structure
- What produces the pattern. Acting here is redesigning: pay bonuses on orders shipped, or on bookings spread through the quarter. The spike loses its reason.
- Belief
- What makes the structure seem normal. Acting here changes what people take for granted: test whether customers sign early when offered early terms. The most leverage, and the slowest.
The sort
- March 31 shipment missed: Event. One occurrence, on one date, to one customer: the tip of the iceberg.
- Last-weekend overtime: Event. A single response to a single crunch. It shows on this quarter's costs and nowhere else.
- Late every quarter end: Pattern. Eight quarters in a row, most late shipments fall in the last two weeks. The event is one of many.
- Quarter-end order spikes: Pattern. Bookings in the final two weeks run at three times the weekly average, every quarter.
- Bonus on booked deals: Structure. Sales bonuses count deals booked by the last day of the quarter, so deals are held until then.
- Targets reset quarterly: Structure. Every quarter starts at zero, so a deal closed on April 1 helps less than one closed on March 31.
- Deals close at deadline: Belief. The sales team takes it for granted that customers sign only under a deadline, so pushing at the end looks like good selling.
- Plant absorbs any spike: Belief. Leaders assume the plant will cope, as it always has with overtime, so the cost of the spike never reaches the sales plan.
An example
A maker of commercial kitchen equipment missed a large customer's shipment on March 31, and the operations team did what it always did: expedited the order and paid the plant overtime for the last weekend of the quarter. The vice president of operations had seen the same scramble every quarter and asked the team to go below the event.
The pattern came from eight quarters of shipping data: most late shipments fell in the final two weeks, when bookings ran at three times the weekly average. The structure behind it was in the sales plan. Bonuses counted deals booked by the last day of the quarter, and targets reset to zero on the first day of the next, so deals were held until the deadline. Below that sat two beliefs nobody had written down: customers sign only under a deadline, and the plant can always absorb a spike.
The team changed the bonus to count orders shipped and tested early-signing terms with ten customers. Seven signed early. Within two quarters, quarter-end bookings fell by half.
Why it works
Most fixes are aimed at the latest event because the event is what everyone saw and what the customer complained about. When the same kind of event keeps returning, the cause sits below it, and every fix at the surface leaves that cause in place.
The model makes the descent explicit and gives each level a different kind of action. At the event, a team reacts. At the pattern, it anticipates. At the structure, the incentives, rules, and information flows that produce the pattern, it redesigns. At the belief that makes the structure seem normal, it changes what people take for granted, which is slow and also the change that lasts. Naming the levels side by side shows a team where its fixes have been aimed, which is usually at the top.
The thinking behind it
Peter M. Senge Peter M. Senge, a senior lecturer at MIT Sloan who brought system dynamics to managers, placed systems thinking among five disciplines of a learning organization in The Fifth Discipline (1990). The book's levels of explanation look past events to the structures that produce them, and its appendix names the system archetypes, recurring feedback structures such as fixes that fail and shifting the burden. The quarter-end scramble is one of these: each expedite fixes the event and leaves its cause in place.
Michael Goodman Michael Goodman, a systems thinking practitioner at Innovation Associates Organizational Learning, taught the iceberg in his article Systems Thinking: What, Why, When, Where, and How? (1997) in The Systems Thinker. Events sit at the surface, patterns of behavior over time lie beneath them, and the structures, including mental models, that produce both sit deeper still. Lasting fixes, he argued, come from changing the structure. The four-level form used here sets the mental models apart as the deepest level.
How to run it
- Name one specific event. The March 31 shipment, not late shipments in general.
- Find the pattern it belongs to. Pull a few years of records and see whether the event is one of many, and when they cluster.
- Ask what produces the pattern. Look for incentives, rules, schedules, and information flows that line up with it.
- Ask what makes the structure seem normal. Write down the beliefs people would give if asked to defend it.
- Say what acting at each level would change. Expedite, staff for the spike, change the bonus, test the belief: one line each.
- Act at the deepest level within reach. Pair it with a surface fix so this quarter's customers are still served.
- Judge it by the pattern. Success is the spike shrinking over several quarters, not one good month.
Where it goes wrong
The session stays at the surface. Every action on the list concerns this quarter: expedite, overtime, a call to the customer. Spot it when no action would change next year's numbers, and add at least one at the structure.
Beliefs turn into accusations. Naming a mental model becomes a charge against the sales team, which then defends it. Write each belief as what anyone in that job would reasonably think, and test it with data.
The deepest level is out of reach. A team picks a belief held three levels above it and stalls. Act at the deepest level the team can actually change, often the structure, and report the belief to those who own it.
When to use something else
- Causal Loop Diagram. When the structure needs drawing as loops of cause and effect, with delays, so the team can see how it feeds itself.
- Leverage Points. When the question is which of many possible changes to a system will move it most.
- Causal Layered Analysis. When the deeper layers are worldviews and shared stories about the future, as in strategy and foresight work.
Where it sits in SOLVE eX
Go to the source
- The Fifth Discipline, 1990
- Systems Thinking: What, Why, When, Where, and How?, The Systems Thinker, 1997