Kumu Blog

Tools and practices for tackling complex systems.

Where to intervene in a system

Last week we introduced a framework for determining whether the levers you are considering have a realistic shot at moving a system. This week we dive deeper into two frameworks from leading practitioners that help you choose among possible leverage points to pursue.

Building shared clarity with Scott Spann

Scott Spann's approach is organized around a few fundamental questions:

  • What's the state of reality?
  • What's the structure that causes that reality?
  • Where do we intervene in that structure/reality?

Scott answers the third question of "where to intervene" by applying 6 different lenses to each of the factors in the map:

  • Levels
  • Levers
  • Loops
  • Micmac
  • Trends
  • Archetypes

For each of these lenses, a factor can essentially score a point based on whether it meets the given criteria:

Levels. Would strategies affecting a given factor rank highly in Donella Meadows' Places to Intervene? Traditionally a factor would need to score in the top 6 of Donella's list to get a point on this criteria.

Levers. Does a given factor affect many other factors, yet is only affected by a few? If something affects 12 things downstream but is only affected by 2, that's a high leverage factor. Not so if it is affected by 10 things - meaning you'll likely have to shift many or all of those to get that lever to move.

Loops. Is the factor part of a reinforcing loop that if primed, would take off and continue positive changes on its own? And do we actually think we have a shot at initiating the reinforcing loop?

Micmac. Does the factor have high leverage and low exposure? Micmac is a cross impact matrix analysis that looks at leverage over time (by taking the logic of the "levers" lens and running it through multiple iterations of the map). We'll save the details of this analysis for a later post, but know that you can end up in one of four quadrants.

Trends. Is the factor trending favorably towards the goal? Ask "How has this factor been changing over time relative to our overall goal?" to see whether it has been increasing, decreasing, or relatively flat. Then ask "For us to make significant progress towards our overall goal, how does this factor need to change in the future?" to see whether people think it needs to significantly increase, increase, remain flat, decrease, or significantly decrease.

Archetypes. Is the factor high leverage based on being part of a known archetype? Archetypes are recurring patterns that we see across varying system types. Consider a riptide: if you're caught in a riptide and know you're in one, you can easily get out of it by following common guidance (swimming with the flow and then out to the side). If you don't know you're in one and you swim against the flow, you might just drown.

Once you've run this analysis for all the factors, total the counts for each one and see which factors scored the highest. It's important to compare these results to your team's intuitions about what the levers are and see how they match up. A disconnect might be an insight to a counterintuitive lever or an indication that your map needs further tweaking.

Here's Scott walking through a live explanation of this approach from a systems analysis done around early childhood systems in Hawaii:

For more on Scott's approach, visit Innate Strategies.

Looking at structural, attitudinal, and transactional drivers with Rob Ricigliano

Rob Ricigliano developed a valuable sorting frame for the different types of factors that are involved in any complex system. These three categories end up proving useful when looking for leverage as well:

  • Structural. The systems (e.g. rule of law, economy) and institutions designed to meet people’s basic human needs.
  • Attitudinal. The attitudes, norms, and intergroup relationships that affect the level of cooperation between groups/people.
  • Transactional. The processes and skills used by key people to manage conflict, solve problems, and turn ideas into action.

The gist of using this frame is that a robust analysis (and map) should have factors from all three categories. If you're overweighted in a given category, it might be a sign that you need to go back to the drawing board or bring different perspectives into your mapping process.

When you're identifying points of leverage, you also want to make sure you've targeted changes that affect structural, attitudinal AND transactional changes and that you are sequencing them in such a way that they align with the natural time frames for each category. Think of the time frames in these buckets:

  • Slow. Transactional changes take time but compared to structural and attitudinal, they are relatively easy to make. Training a key stakeholder to adopt a new behavior or skill is a finite, achievable task.
  • Slower. Structural changes are between transactional and attitudinal change. Something like policy change has a long cycle, but once the change is enacted the consequences happen rather quickly.
  • Slowest. Attitudinal change is the slowest. Changing long held beliefs among communities (especially when those are negative and certain groups feel wronged) can take decades.

Once you've done this, there are a handful of additional questions that can prove useful in figuring out where to engage a system:

  • Where are factors in flux? Systems resist change. When you find areas that are already moving, it's a sign that you might be able to move the area significantly.
  • Where are the bright spots in the system? Are there surprising areas of positive deviance? What might these hint at in terms of creative ways to intervene?
  • Where are the attractors? Are there certain factors that have a surprising "gravitational pull"? Are there common states a system seems to end up back in regardless of the intervention?
  • Are there factors that have opposite impacts in different loops? When one factor has an opposite impact in two loops, a powerful intervention can be to amplify the positive impact while finding a way to dampen, or eliminate the negative impact.
  • Are there factors with higher potential for ripple effects? Which factors, if changed, would ripple throughout the system based on causal links and reinforcing loops?
  • What strikes you as especially prevalent? Which factors are you seeing appear over and over again, across loops, conversations, and analyses?
  • Which factors don't show up everywhere, but are especially powerful when they are present? Are there factors that are rarely mentioned, but when you see them are suprisingly powerful?


We've presented two frameworks for identifying leverage points in a system. As you might have expected, there's no clear cut way to know the best way to engage a system. But by being thoughtful in your approach, engaging the system and actively listening for feedback from the system, you can iterate your way to effective interventions.

We'll leave you with a great takeaway from Scott. As you're working to determine where to intervene, you'll probably be confronted with both cynics and idealists. As a leader of systems change, you have a unique role:

The reality is that neither cynicism nor idealism is going to get the job done. The role of the leader is to internalize both of these potentials in their nervous system and to use that gap as a source of creative tension out of which innovation can arise.
- Scott Spann

Subscribe for updates

Loading comments...