• Clair Whitmer

When Data Takes the Wheel

Updated: May 22, 2019

I had an interesting conversation with a former colleague the other day about when a Blue Cigar is just a Blue Cigar.He was on assignment on a piece for IBM about how creative teams use data and, knowing that I now work on these kinds of teams, he contacted me as a source.


The inspiration for the story was Douglas Bowman, Google’s first visual designer. Bowman left Google in 2009, in part because he was tired of testing shades of blue. My friend was re-examining this question of how much influence data should have on design. In his 2009 public good-bye letter, Bowman wrote:

I won’t miss a design philosophy that lives or dies strictly by the sword of data. Douglas Bowman

Fair enough. But that doesn’t mean data is the wrong weapon for Google. Google makes money on literally every click so having a teeny-tiny difference in click-rate based on the shade of blue could — theoretically — translate into measurable revenue. That would be true for only a handful of businesses in the world, maybe even Google alone.


For most companies, spending endless cycles A/B testing shades of blue would just be operational money down the drain without a tangible return. So it’s not so much that Google is wrong as that working for Google’s design team sounds boring.

The four stages of Data Dependency

The trick for most product teams is to make intentional choices about where design merits this kind of scrutiny; I normally wince at the expression “move the needle”, but, yes, where can design move the revenue needle, where design consistency should be used to influence user behavior, and where design needs to demonstrate creativity on behalf of the brand.


I try to adhere to the axiom that every pixel of information on the screen should convey meaning; I (discretely) roll my eyeballs at designs that are trying to be creative for creativity’s sake. Establish design rules, test them to see if your intended meaning is coming across, and break them only when you’re trying to convey new information.


But what happens if a brand is striving to be perceived as “innovative” or “disruptive”? In this case, breaking the mold could be the element that communicates the brand message. But how to test that?


I’m thinking a lot about this question right now because I’m starting a new project where I need to use our brand to inspire innovation from our users. Is this one of those times where design can be used to do that? Or do you count on the content to do so? Either way, how do I know if it’s working — or just creating a hot mess?


One of the other people cited in my friend's piece was Marc Engelsman, Vice President of Strategy and Analytics at Digital Brand Expressions, a digital marketing agency in Princeton, New Jersey. Englesman was quoted as saying that:

“Data should be used to influence decision-making, not do the decision-making.” Marc Engelsman

That sounds right. My data-indifferent CEO would certainly agree.


But this is the Catch-22 part: once you decide that a subjective judgment — let’s say, for example, from your CEO — is the right design decision, it’s very hard to reverse engineer that.


You can survey users after the fact e.g. “When you’re using this product, do you think of the word ‘innovation’ or the expression ‘warmed-over rehash’? Does this product inspire you to invent a better mousetrap or to watch Fantasy Island reruns while eating potato chips?”


But if you released it that way because your boss liked it and then the product isn’t performing, the design may well get a pass because you already accepted subjectivity as a legitimate measuring stick. It reminds me of a story about staffers telling Bill Clinton that polling showed that the American people didn’t particularly like his wife. His response:

“Oh, it’s just her hair they don’t like.”

This is how it goes in so many post-launch conversations with stakeholders: they’ll blame any part of a product but the parts that they personally evangelized.


This is why creatives like data: it helps them avoid this conversation.

#datascience #analytics


Text: 510-731-7890

© 2019 by Clair Whitmer

Find me

  • linkedin-square
  • Twitter Square