Looking sideways at the design squiggle

Andy Matuschak
Khan Academy Early Product Development
3 min readAug 29, 2018


Though everyone seems to draw it slightly differently, the coarse shape of this diagram shows up ubiquitously when designers describe their process (including us!):

The Process of Design from a great height. Damien Newman.

This squiggle beautifully captures the nonlinear shift from early messiness to later neatness, but—almost with a wink—it willfully obscures what’s happening during those up-and-down loops. As I’ve tried to describe our process to others, I’ve realized we can see a bit more structure through the zigs and zags:

Just turn the squiggle like a top until we’re looking down into its oscillating center—now it’s a spiral! While the contours do meander, each turn of our spiral tends to trace through the same fundamental regions:

(the specific activities marked along the spiral are just illustrative examples, not prescriptions)
  • We center on questions, which we refine as best we can.
  • We gather all that surrounds those questions through research.
  • We generate approaches through ideation.
  • We test the questions, the research, and the ideation
  • Then we repeat.

Our team articulates our top questions each week, but we don’t formalize these phases through heavy structure: we naturally proceed through them from the squiggle’s messy start through to its tamed tail.

Of course, the activities in those quadrants look different as the project evolves. For instance, the research phase initially paints broad strokes: we’ll conduct interviews without specific hypotheses; we’ll read scholarly reviews and surveys; we’ll browse offerings in the space. On subsequent turns, we’ll have more specific questions: we’ll conduct more targeted interviews; we’ll read literature from specific authors and keywords; we’ll dig into specific players we heard come up. Similarly, the testing phase might start with paper prototypes or marketing copy—and end up with fully-usable implementations.

This cycle extends beyond the initial design work for a project. Even once the squiggle has calmed down, if we turn it on its side, we’ll see gentle loops through these phases as the production team iterates on the concept. Research may manifest as “reading support tickets”; ideation may manifest as proposing alterations of existing designs.

I was intrigued to find a similar framework articulated in defense software development by Barry Boehm in 1986:

Boehm, B. W. (1988). A spiral model of software development and enhancement. Computer, 21(5), 61–72. Figure 2.

This is a much more regimented version of the spiral illustrated above, meant to be executed in a context with contractually-mandated timelines and perhaps less generativity. It does add a few interesting notions:

  • The left portion of the x axis represents the “commitment partition”: a new loop of the spiral requires an explicit decision to proceed.
  • This model is acutely interested in risks, which are important but which are only one kind of question we end up refining and asking in our first phase. Often the core question for one spiral loop is something more exploratory, like “what types of activities might best help students reflect on what they’ve learned and consolidate the material?”
  • Here the radial dimension represents the cumulative cost of the project. In our work, this is much less consistent: in earlier phases, research is often more expensive (many hours interviewing teachers!); in later phases, ideation is often more expensive (building prototypes!).
  • The detailed activities marked along this spiral aren’t examples of what might be done at that phase: they’re meant to precisely specify the activities there. We’ve not found that level of enforced structure constructive in investigative design research.

I wonder how these models might converge! Are some of the activities in our research spiral more consistent than others? Perhaps over time we’ll notice more regularity we can use as handholds along the way.

On a personal level, at least, this type of loose consistency has been a helpful lighthouse in the sometimes-enigmatic seas of research.



Andy Matuschak
Khan Academy Early Product Development

Wonder, blunder, salve, solve! Exploring empowering future possibilities in education with team at Khan Academy, where I lead Early Product Development.