Designing Systems at Scale

Designing Systems at Scale
Alëna Iouguina, Senior Lead Systems Research, Shopify

From the Design Operations Summit website:
To be alive in the twenty-first century is to rely on countless complex systems that profoundly affect our lives, where small mistakes can turn into massive inefficiencies and failures. These failures seem to stem from very different problems, but their underlying causes turn out to be surprisingly similar.  This talk will cover about the common threads of these failures, how they can be detected and ultimately prevented using various research disciplines – from market research to anthropology to data science. We need to get good at designing and building for systems to tackle multi-dimensional problems at the intersection of people and technology.

She came here 20 years ago from Russia; she studied at the Fashion Institute of Technology with Judy Ellis and her magical toy design program.


Judy taught about the importance of aesthetics, manufacturing, and having a clear underestanding of the brand.  But Judy also brough something else elusive – she taught that tools and processes permit us to make magic happen.  In fact, the toy is a vehicle to achieve the real goal – the art of play.

Alëna moved to Canada and began to focus on engineering.  One of her professor’s recommendations was to save time by using existing engineering frameworks.  After some reflection, she made the decision to quit engineering.

More recently, she has come back to design with a new appreciation for both.  How to enable people to create their own possibilities.  We need a well laid out process (in engineering spirit) but must be transparent (in the design spirit).

When she joined Shopify in Ottawa, she was one of four researchers.  They are now fifty-four researchers, and each has it’s own target audience and product line.  But the decisions of one product lins affects the others, and there are risks if the alignment slips.  They are using their knowledge of their audiences to create procucts.  But how to ensure that knowledge is well gathered and shared at the right time?  They wanted to have the right knowledge available in a centralized way.

They share user research, market insights, data science, and findings from support teams.  Historically, there has been more value placed on knowledge creation rather than knowledge re-use. Knowledge needs to be thought of as a circular process:


Create.  We need a diversity of data to support how we make decisions as humans.  02b-ioouguinaThe brain is able to surface the right information at the right time, which we gather from our senses.  That gets relegated to the what stream.   Meanwhile the brain is focused on the where in parallel.  It gathers all kinds of spatial information.  The what stream allows us to evaluate the sensory aspects.  And the where helps provide context.

What we know from research on the brain is that these data streams work well together; object-related data and context-related data.  The two streams merge in the hippocampus, where connections are formed.  The anterior and posterior hippocampus have different specialities:03-iouguina.png

It’s the context-related data make their way to decision-making centers first.  After the knowledge is generated, the brain then assesses validity.  That is why we want to have exploratory research before evaluative research – so we can make great decisions.  This knowledge creation phase optimizes for the next phase.

Distribution.  Rather than managing the knowledge itself, manage the environment where it takes place.  Researchers move across product lines, which enables and requires continuous knowledge exchange.  The data scientists operate the same way.

Apply.  How do we bring this to the business?  This is where the art of play comes in.  We have talked about diverse streams of data.  We need to go further to undrestand context, and find patterns so we can make the right kinds of change.  Our analysis turns knowledge into action.

Evaluate.  The final step flows out of the first three.  Each time the product team cycles through this process, they move from KPIs to wisdom, and enables them to react more quickly to changes over time.

Some key mechanisms for them have included supportive leadership, asset visibility (e.g. their Frustrations system), and knowledge management.  They have evolved their onboarding process from a spreadsheet into Trello based on the informal learnings and comments in their original document.

Alëna shared three big takeaways with us:

  • Companies today are complex systems split up by functions, and aligned by information flows; they reflect how our brains operate.
  • Research has a major role to play in this biologically informed process.
  • And finally, that elusive element (with thanks to Judy Ellis): “With proper maintenance, knowledge becomes a vehicle that probes into the most hidden spaces of what’s possible and brings to the surface the best decisions.”

1 Comments on “Designing Systems at Scale”

  1. Pingback: Design Ops 2018 – Recap | Natalie Hanson

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: