Here is the video for this talk:

How to craft a data story

Anita Lillie

September 25, 2014

Data from this talk:

Who are we?


Where we live

Mountain View

I live in

What we like to do



"making a zine"




Happiest age

Who am I?

  • Previously: LinkedIn, Nokia, MIT Media Lab, Stanford
  • Music, social networks, genetics, healthcare
  • Data Scientist at Practice Fusion

I design and develop data interfaces.

What I'm not going to talk about



Big Data


Mountain biking

Previous work

Venn diagrams


♥ Phil Collins ♥

What I am going to talk about

  • How to think about data stories

  • How to have empathy for your user

  • The creation process

  • Tests to check your work

Where I find inspiration...

The bulk aisle

Footsteps after a puddle

Wear on door handles



Wear on doors

Football fields


Fingerprints on an iPad

Carnegie Mellon fence

Carnegie Mellon fence

Carnegie Mellon fence

Evidence of something interacting with something else

...leaving its mark.

All of these examples are

Why physical metaphors?

  • Human cognition is built for these physical things.  We see these kinds of patterns easily.
  • Thinking in this way helps you choose visual mappings that make sense.
  • Making something mimic the real world gives it meaning.


backwards-engineering naturally-occurring visualizations.

  • You will ask the right questions.
  • You will come up with better data stories.
  • You will tell them in a more intuitive way.

My Process

My Process

  1. Define my audience
  2. Understand the data
  3. Lots of ideas for questions and messages
  4. Explore the data (...more ideas)
  5. Plan out possible stories
  6. Select a few key messages to tell one story
  7. Overall story design
  8. Make it pretty

Iterate and test constantly.


  1. Be honest. Be direct.
  2. Keep it simple
  3. Remember your audience
  4. Use vis for a reason
  5. Make it real and relate-able
  6. Explore the real data early
  7. Show appropriate context
  8. Make it pretty at the end
  9. Iterate!

Testing your work

Squint test

Squint at your work so you can't read any of the words.  What patterns do you see in the data?  What questions do you ask?


If you don't see any patterns, it's not right yet.

The test doesn't always apply...

10-second test

Pretend to wipe your brain clean. Or grab someone new.


Look at your work for only 10 seconds. What messages do you see?

What's your user's

Next Question?

Wipe your brain clean again. Or grab a new person.

Look at the vis for long enough to get an understanding of it.  What is the next question you ask?

Is it related to your story?  Is it a question born of curiosity rather than confusion? Does it reflect a new perspective?

Our data

Our data

What stories could we tell?

Explore this data yourself!

Survey data:​

Hobbies data:​

Contact me



  • Exploring data:
    • Tableau
    • R
    • Excel
  • Design
    • Illustrator
  • Front-end programming
    • JavaScript (D3)

How to craft a data story

By Anita Lillie

How to craft a data story

Watch video of the full talk:

  • 6,915