Developed by Chris Argyris and Donald Schön (Theory in Practice, 1974).
Separates fixing actions within current assumptions (single loop) from questioning the assumptions and goals themselves (double loop), and builds the conditions for the second.
Eight responses to a sales miss, sorted by which loop they run
Each card is a response from the packaging maker's review, after three missed quarters. One is placed as an example: select each other card, then the box for what it changes, the action or the assumption behind it. Show the sort to compare yours. Illustrative.
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
- Not placedPlacedSingle loopDouble loop
Proposed in the review
What the response changes
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.
- Single loop
- Adjusts the action and keeps the goal and its assumptions. Right when the plan is sound and execution slipped; useless when the same miss keeps repeating.
- Double loop
- Questions the goal or a belief the actions depend on. Needed when harder work keeps producing the same miss.
The sort
- 30% more calls: Single loop. More calls to the same plant managers: the action grows, and the belief that they decide is untouched.
- Bigger discount: Single loop. A cheaper version of the same offer to the same buyer.
- Ask who signs: Double loop. Tests the belief that plant managers decide. Fourteen of twenty deals were decided by a parent company's purchasing group.
- Retrain the reps: Single loop. Sharper answers to the objections plant managers raise, whoever turns out to be deciding.
- Doubt the pitch: Double loop. Tests the belief that machines sell on specifications. Purchasing groups buy on total cost.
- Two more reps: Single loop. More people running the same plan in the same territories.
- Redraw the map: Double loop. The map encodes the plant-manager belief. Redrawing it around parent companies changes the assumption, not only the effort.
- Filter leads harder: Single loop. Better filtering of the same kind of lead against the same buyer profile.
An example
A maker of packaging machines had missed its sales target three quarters running, each time by about a quarter. Every review produced the same list: more calls, a deeper discount, objection training, two more reps. The team did all of it, worked longer, and missed again.
The new sales director split the next review in two. First she wrote down what the plan assumed: the buyer is the plant manager at a mid-size food plant, and a machine sells on its spec sheet. Then she asked the team to test each assumption against the last twenty lost deals. Fourteen had been decided by a purchasing group at a parent company, after an acquisition, and none of those buyers had seen a spec sheet. Nobody had raised it, because questioning the plant-manager model meant questioning the territory map, and the territory map was the vice president's.
The team changed the assumption, rebuilt three territories around parent companies, and wrote a total-cost case for purchasing groups.
Why it works
Every action rests on assumptions about what the goal is and how the world responds. When results miss, the natural move is to adjust the action and leave the assumptions alone, because they feel like facts and questioning them threatens the people who set them. That works while the assumptions hold. Once they stop holding, effort only produces the same miss faster.
The tool's value is the second question it forces: is the miss a sign that the action was weak, or that the theory behind it is wrong? Writing the assumptions down makes them testable. Asking what nobody is allowed to say finds the assumptions that are protected by rank or habit, which are usually the ones most worth testing.
The thinking behind it
Chris Argyris Chris Argyris, a Harvard scholar of organizational learning, drew the distinction with Donald Schön in Theory in Practice (1974). Single-loop learning corrects an action against a goal; double-loop learning questions the goal and the assumptions that govern the action. He also separated the theories people say they follow from the ones their behavior shows, and in Overcoming Organizational Defenses (1990) turned to the defensive routines that keep professionals from examining either.
Donald Schön Donald Schön, an MIT philosopher of professional practice, was Argyris's coauthor on Theory in Practice (1974) and Organizational Learning: A Theory of Action Perspective (1978). In his own book The Reflective Practitioner (1983) he described how skilled professionals think in the middle of their work and afterward, the habit double-loop learning depends on, and in The Design Studio (1985) he took the architecture studio critique as a model of it.
How to run it
- Write the governing assumptions down. List the goal and the beliefs about customers, markets, and the team that the current plan depends on.
- Look at what actually happened. Gather the results, including the side effects and the pattern across several cycles, not only the latest miss.
- Ask the single-loop question. Name what could be adjusted within the current assumptions, and keep that list.
- Ask the double-loop question. For each assumption, ask what evidence would show it is wrong, and go and look for that evidence.
- Name what nobody raises. Ask which topics the team avoids, especially those tied to a senior person's past decision.
- Decide which loop the problem needs. If adjustments have already been tried and the miss repeats, treat it as an assumption problem.
- Change the assumption on the record. Write the revised assumption into the plan, so the next review tests the new one.
Where it goes wrong
The review asks the double-loop question and answers it single-loop. A team asks whether its segment is right, agrees that it is within ten minutes, and returns to the call targets. Ask for evidence that would show the assumption wrong, and look for it before answering.
Everything gets questioned. A team that reopens its strategy after one bad quarter never executes anything long enough to learn from it. Try the adjustments first; move to the assumptions when the same miss repeats.
The undiscussable stays undiscussed. If the assumption most worth testing belongs to the most senior person in the room, the conversation will route around it. The senior person has to name it first, or it will not be named.
When to use something else
- After-Action Review. When the task is a single event and the team needs a quick, regular comparison of what was planned with what happened.
- Assumption Audit. When a new plan's assumptions need to be listed and ranked by risk before anyone executes it.
- Iceberg Model. When a recurring problem needs to be traced from events to patterns to the structures and beliefs that produce them.
Where it sits in SOLVE eX
Go to the source
- Theory in Practice: Increasing Professional Effectiveness, 1974
- Teaching Smart People How to Learn, Harvard Business Review, 1991