Cynefin

Thinking Tools1.3 Framing the SituationAlso known as Cynefin Framework

Developed by Dave Snowden at IBM (Cynefin, a Sense of Time and Place, 2000), set out in full with Cynthia F. Kurtz (The New Dynamics of Strategy, 2003), and brought to managers with Mary E. Boone (A Leader's Framework for Decision Making, 2007).

Sorts a situation by how cause and effect behave (clear, complicated, complex, chaotic, or confused) and matches the way of deciding to the domain.

Seven problems from one operations meeting, sorted by how cause and effect behave

Select a problem, then the domain it belongs in: the domain lights and says how to decide there. Confusion runs down the middle, for what nobody can place yet. Show the sort to compare yours. Illustrative.

Items (select one, then a box)
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion
  • Not placedPlacedComplexComplicatedClearChaoticConfusion

Room to study first

Study firstAct now
Complex
Confusion
Complicated
Chaotic
Clear
Seen only afterNot yet agreedKnowable in advance

Cause and effect

One paycheck short.Not placed yet.One paycheck short.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.One paycheck short.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.One paycheck short.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.The cause is known and the fix is written down: payroll runs the standard correction today.One paycheck short.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.One paycheck short.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Card data leaking now.Not placed yet.Card data leaking now.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.Card data leaking now.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.Card data leaking now.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.Card data leaking now.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.Card data is leaking now. Take checkout offline within the hour; the root cause can wait until the bleeding stops.Card data leaking now.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Recurring line jam.Not placed yet.Recurring line jam.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.Recurring line jam.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.There is a cause, and a maintenance engineer with the sensor logs can find it.Recurring line jam.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.Recurring line jam.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.Recurring line jam.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Choosing an ERP vendor.Not placed yet.Choosing an ERP vendor.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.Choosing an ERP vendor.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.Hard, but knowable: specialists can test the options against the requirements.Choosing an ERP vendor.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.Choosing an ERP vendor.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.Choosing an ERP vendor.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Selling in a new country.Not placed yet.Selling in a new country.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.No study predicts how customers there will respond. Run three small launches with a date to read them.Selling in a new country.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.Selling in a new country.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.Selling in a new country.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.Selling in a new country.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Four-day week for churn.Not placed yet.Four-day week for churn.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.Staff, managers, and customers will all adapt to it. Trial it in two teams and watch what changes.Four-day week for churn.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.Four-day week for churn.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.Four-day week for churn.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.Four-day week for churn.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.VPs split on the problem.Not placed yet.VPs split on the problem.Complex: Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.VPs split on the problem.Complicated: There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.VPs split on the problem.Clear: Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.VPs split on the problem.Chaotic: No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.VPs split on the problem.Confusion: Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.Sales sees a pricing problem, operations a delivery problem. Separate the two claims and place each one.

Your sort matched 0 of 7.Your sort matched 1 of 7.Your sort matched 2 of 7.Your sort matched 3 of 7.Your sort matched 4 of 7.Your sort matched 5 of 7.Your sort matched 6 of 7.Your sort matched 7 of 7.

Complex
Cause and effect show only in hindsight. Probe, sense, respond: run small trials that are safe to fail, and grow what works.
Complicated
There is a right answer, and finding it takes expertise. Sense, analyze, respond: bring in the expert and the data.
Clear
Cause and effect are plain to anyone. Sense, categorize, respond: apply the known practice and move on.
Chaotic
No visible link between cause and effect, and harm is spreading. Act, sense, respond: stabilize first, then work out what happened.
Confusion
Nobody agrees which kind of situation this is. Split it into parts and place each part on its own.

The sort

  • One paycheck short: Clear. The cause is known and the fix is written down: payroll runs the standard correction today.
  • Card data leaking now: Chaotic. Card data is leaking now. Take checkout offline within the hour; the root cause can wait until the bleeding stops.
  • Recurring line jam: Complicated. There is a cause, and a maintenance engineer with the sensor logs can find it.
  • Choosing an ERP vendor: Complicated. Hard, but knowable: specialists can test the options against the requirements.
  • Selling in a new country: Complex. No study predicts how customers there will respond. Run three small launches with a date to read them.
  • Four-day week for churn: Complex. Staff, managers, and customers will all adapt to it. Trial it in two teams and watch what changes.
  • VPs split on the problem: Confusion. Sales sees a pricing problem, operations a delivery problem. Separate the two claims and place each one.

An example

The chief operating officer of an online retailer ran one weekly issues meeting, and every issue got the same treatment: an owner, a root-cause analysis, and a recommendation in two weeks. One week the list held seven items, among them a short paycheck, a packing line that jammed every third shift, a checkout outage that had exposed card data, a plan to sell in a second country, and a proposal for a four-day week to cut staff turnover.

She sorted them by how cause and effect behaved. The paycheck went to payroll with the standard correction. The line jam went to the maintenance engineer with the sensor logs. The checkout outage could not wait two weeks: the site went offline within the hour, and the analysis came after. The new country and the four-day week got no analysis at all. Each got small trials with a date to read the results, because no study could predict how customers or staff would respond.

Why it works

Most management methods assume that study before action finds the answer. That holds when cause and effect are stable, and fails in two directions. In a complex situation (people, markets, cultures) the system changes as people act on it, so analysis produces confident forecasts that turn out wrong, and only trying things shows what works. In a chaotic one, waiting for analysis lets the damage spread.

Naming the domain first stops a team from using a good method in the wrong place: best practice applied to a market entry, or a six-week study of a breach. It also makes disagreement useful. Two people arguing about how to proceed often disagree about which kind of situation they face, and that can be settled first.

The thinking behind it

Dave Snowden Dave Snowden began the Cynefin framework in 1999 at IBM Global Services and first set it out in Cynefin: A Sense of Time and Space (2000), a chapter in the collection Knowledge Horizons. It sorts situations into domains by how cause and effect relate (clear, complicated, complex, and chaotic, with confusion between them) and gives each its own way of acting, the best known being safe-to-fail experiments that probe the complex domain.

Cynthia F. Kurtz Cynthia F. Kurtz worked with Snowden at the IBM Cynefin Center and co-wrote The New Dynamics of Strategy: Sense-Making in a Complex and Complicated World (2003) in the IBM Systems Journal, the first full published statement of the framework. As its title says, the paper set the complex apart from the complicated. She later developed participatory narrative inquiry and wrote Working with Stories.

Mary E. Boone Mary E. Boone, a management consultant and president of Boone Associates, wrote A Leader's Framework for Decision Making (2007) with Snowden in the Harvard Business Review, the article that brought Cynefin to managers. It named the domains simple, complicated, complex, chaotic, and disorder, and set out how a leader's way of deciding has to change from one domain to the next.

How to run it

  1. List the situations on the table. Write each as one concrete problem; a bundle such as "operations issues" cannot be placed.
  2. Ask how cause and effect behave. Is the answer known, knowable by an expert, visible only in hindsight, or not visible at all?
  3. Put anything disputed in confusion. If the group cannot agree, split the situation into parts and place each part.
  4. Match the method to the domain. Known practice for clear, experts for complicated, small trials for complex, and fast action to stabilize for chaotic.
  5. Name what would move it. Write the signal that would shift each item, such as a trial result or the end of an outage.
  6. Revisit the placement each cycle. A crisis settles into complex or complicated work once stable, and a clear routine can tip into chaos when conditions change.

Where it goes wrong

Everything lands in complex. The label sounds sophisticated and excuses a team from finding answers an expert could give. Ask who would know: if a named person could solve it with time and data, it is complicated.

Clear is treated as safe. Routine work run by procedure for years breeds complacency, and when conditions shift unnoticed, a clear situation can fall straight into chaos. Watch the routines that throw the most exceptions.

The sort changes nothing. Teams place every item and then run them all the way they always did. The placement is worth something only if the method changes: a trial plan for each complex item, a named expert for each complicated one.

When to use something else

  • OODA Loop. When the situation is chaotic or fast-moving and the team needs a cycle for acting and reorienting under pressure.
  • Randomized Controlled Trial. When an item sits in complex and a probe needs a controlled test to read its result.
  • Issue Tree. When the problem is complicated and the work is breaking it into parts an analyst can test.

Where it sits in SOLVE eX

Go to the source

  • Cynefin, a Sense of Time and Place, Knowledge Horizons, Butterworth-Heinemann, 2000
  • The New Dynamics of Strategy: Sense-Making in a Complex and Complicated World, IBM Systems Journal 42(3), 2003
  • A Leader's Framework for Decision Making, Harvard Business Review, 2007

Back to the catalog