Designing at leverage points

How to find the right spot in any system to create maximum impact.

Adam Saint
Shopify UX
Published in
8 min readApr 3, 2017

--

“For want of a nail the shoe was lost,
for want of a shoe the horse was lost,
for want of a horse the knight was lost,
for want of a knight the battle was lost,
for want of a battle the kingdom was lost.
So a kingdom was lost — all for want of a nail.”
—13th century Medieval proverb

Designers grapple with mounting complexity every day as we attempt to tease order out of chaos. Our profession is tasked with designing and building unbelievably complex systems. We lack the necessary language and tools to dismantle these systems, understand their components, and find the right areas to focus our work.

Our field tends to evangelize the magic of holistic thinking, but we’ve become complacent in our methods. Our oversimplified models don’t accurately resemble the problems we’re trying to solve. As a trade, designers (in the broadest sense of the word) have long labored at the margins of influence in most companies. Too often relegated to the role of a support function, brought in to ‘make it nice’ during a polish phase at the end of a project.

The scope of our effectiveness has been limited where it counts, because of our reluctance to fully embrace the principles of cause and effect. We’ve yet to transition into the next era of our profession, in which we become the caretakers and gardeners for systems. We’re due for an overhaul of our core methodologies.

Tools of another trade

Fortunately, we have a wealth of wisdom available to us from more established communities of practice. In this case, in order to realize our objective of understanding cause and effect, Donella Meadows’s work in systems thinking and analysis provides us with a powerful concept for thinking about where and how to devote our attention in a system in order to realize our objectives.

Meadows’s central insight was that in any system (e.g. cities, economies, ecosystems, and so on), there exist levers where a “small shift in one thing can produce big changes in everything”. She went on to outline twelve leverage points that can be used to trigger these changes, with each successive level giving you a greater magnitude of leverage over the state of the system. For brevity’s sake, I’ve visualized the model here with a diagram:

Donella Meadows’s twelve leverage points to intervene in a system. You’re probably thinking that the descending order of leverage is confusing, which it is, but it’s also true to the source.

There’s more to it than what’s pictured here, and I would encourage you to check out the source material over at the Donella Meadows Project. Her commentary and use of metaphor to explain the mechanics of each leverage point is enlightening and fun, especially if you’ve never spent much time pondering how bathtubs actually work.

Pulling levers

At this point you’ve already started to unpack the concept in your mind and realized that it’s not much of a leap to adapt something like this into the design process. You could use Meadows’s model off-the-shelf to think more clearly about your work, but if we bend the concept to our purposes a little bit, a useful structure starts to emerge.

I propose that we adopt a working vocabulary for designing at leverage points. This isn’t a literal representation of every scenario, but is detailed enough to give us some visibility on the leverage points that are available in most of the systems we’ll be working with.

  1. Parameters and representations (least leverage): aesthetics, interfaces, layouts, surface language, etc.
    (Approaches: Interface design, industrial design, typography, copywriting)
  2. Flows and feedback loops: adjusting the user’s route from point A to B, using positive and negative feedback loops to alter behaviours en masse.(Approaches: UI flow diagramming, Dan Saffer’s Microinteractions, Nir Eyal’s Hook model)
  3. Structure, meaning, and availability of information: redefining nomenclature, changing what information is available, how much of it there is, and who has access to it.
    (Approaches: The entire fields of information architecture and content strategy)
  4. Goals of the system: changing your definition of success or strategy, altering your KPIs and metrics, catering to a new user archetype.
    (Approaches: OKRs, Clay Christensen’s Jobs to Be Done Theory, Google Venture’s Goals-Signals-Metrics process)
  5. Principles and values of the greater context (most leverage): What’s the next largest context? A product lives inside a company, which exists in a market, and so on. If you can shift the nature of this larger context, you can gain immense leverage.
    (Approaches: You’re not going to find answers to this in a Medium article. This is one of the great challenges of our time.)

Here’s another diagram to help us visualize the concept:

If we frame our work with this model, we’ll be able to:

  • Break down and categorize the components of a system to better understand the relationships between all of the parts.
  • Identify opportunities and entry points into systems, determining the best places to intervene.
  • Scope, segment, and understand the problems we’re trying to solve.
  • Realistically assess the level of impact we can expect to see from a change we implement.
  • Communicate more clearly and consistently between designers working on a system.
  • Communicate and frame our work more clearly with other stakeholders working on a system.
  • Help us to decide which methods or tools to use depending on what point of leverage we’re targeting.

You might interpret this as a value judgement on the relative importance of the work being done at each leverage point, but this isn’t the intent. If you’ve been isolated to one layer in this model, it’s unlikely that you’re going to be able to do the kind of work you really should or want to be doing. And I don’t think any designer would seriously contend, for example, that the layout of one screen is equal to changing the total number of screens in an entire system. But we can’t afford to trivialize work at lower leverage points, because everything is connected, and every detail is another nail.

So a kingdom was lost

Once you’re bought into this mode of thinking, an interesting corollary to consider is that designers are uniquely able to affect change in complex systems. This is because we can make real things and wedge them in to a context — we can create new levers.

In his book “Dark Matter and Trojan Horses: A Strategic Design Vocabulary”, Dan Hill describes the process of baking an agenda into a design artifact, or what he calls a trojan horse.

To paraphrase: a trojan horse is a thing that’s made to carry consequences beyond the artifact itself. His example comes from architecture and community planning, where a building is created to explicitly change the legislative fabric of the city where it’s located. The structure itself was conceived in a certain way, with a particular philosophy, with the intent to influence the other buildings, laws, and people living in its midst.

We’ve become pretty comfortable thinking about rules changing our environment, but it can be empowering to think about the act of reshaping our environment to modify the rules. This also aids in dispelling the perception of visuals as a superficial aspect of design. The visible things we create are almost always loaded with invisible consequences. It’s never just an interface, or a piece of copy. It’s never just a nail.

This is where the model gets genuinely interesting. We can begin to plot the things we make along their levels of impact and reason about the interconnected relationships of all of the elements in the system. I’d argue this is the most compelling version of the visualization that we can consider:

This could be a very powerful way of laying out all of the intangible goals of your product or company, intermingled with their tangible outputs. I find these kinds of connection are seldom talked about or represented in business.

Our systems are alive, our processes are dead

It’s worth noting that this idea bears a resemblance to Jesse James Garrett’s classic representation of the elements of user experience, pictured below:

The Elements of User Experience by Jesse James Garrett c. March 2000

This makes perfect sense. All design work falls somewhere on this ladder of abstraction, from ideas to concrete deliverables. And our ability to navigate up and down the ladder is essential. But, while Garrett’s framework is useful for communicating the general structure and sequencing of a static UX project, the purpose of designing at leverage points is to identify the right place to intervene in a living system (such as a business) to trigger the changes you have in mind.

Complex systems are not segmented into the tidy bento box diagrams we’re used to seeing. They’re messy, entangled latticeworks of context where tugging on one thread pulls a dozen more. We‘ve traditionally used boxes and arrows to represent these concepts, but we’d be better off working with simulations and prototypes that react and distort as data flows through them, revealing the ripples caused by tweaking variables at different leverage points.

Okay, how might we proceed?

Let’s recap and zoom out for an overview of what we’ve covered so far:

  • We work with complex, living systems (products, services, businesses, etc.).
  • Our traditional processes and tools are not well suited to mapping systems and figuring out where to make changes.
  • The concepts and models from systems thinking and analysis are very useful for the kind of work designers are trying to do these days.
  • We’ve always known about layers of abstraction in design, but now we can think about what kinds of leverage and impact are possible in each layer, and then choose the right entry point depending on our goals.
  • We’d be better off reasoning about our work using simulations, or prototypes fed with data.
  • Designers, or anyone who can make tangible things, have the unique ability to make new levers and drive them into a system.

Now when we set out to alter a system, or build one from scratch, we have a means for visualizing and organizing its different pieces, and plotting our best means of causing our desired changes.

This is also just a single entry in the catalogue of models for making sense of complex systems, and by no means a definitive approach. This is my first pass at adapting a concept from a field that has struggled with these challenges for decades. I hope we can use concepts like this to think more clearly and do better work together. I’m certain that these ideas need a few rounds of refinement, and I welcome your feedback.

If this idea proves to be useful for a significant number of people, I’m considering developing this into a series about adapting tools, processes, and models from systems thinking (and other fields) to equip designers with the means to do their best work. Let me know what you think, and thanks for reading.

Much gratitude to Meg Robichaud for the proverbial cover art. These are the kinds of things we think about on Shopify’s UX team. If this makes you curious about what it’s like to work on team of hundreds of interdisciplinary UX practitioners, building unbelievably complex systems, I encourage you to get in touch with me on Twitter!

--

--