Thursday, January 23, 2020

What you learn when you're shot at.

Just yesterday I ran across this picture from a site that features stories about World War II. Along with the picture, came a story of a concept I was unfamiliar with:
Survivor-Ship Bias.

Here's a succinct definition of Survivor-Ship Bias: Drawing conclusions from an incomplete set of data because that data has "survived" some selection criteria.

Here’s the story as I read it.

An American plane came back to England after a bomb-run over Germany. The red dots on the outline above show all the places the bomber was hit by German anti-aircraft fire.

Engineers looked at this diagram and others like it. They said ‘there are no bullet holes on the engine and the cockpit—they don’t need armor-plating. The wings and the tail are where planes get hit. We’ll reinforce those areas.”

That seemed to be wise and the consensus among the engineers. You put extra protection on the places that need it most.

One mathematician on the team, Abraham Wald, analyzing the data looked at it and saw something completely different. The planes the engineers were analyzing had one thing in common. They had successfully returned. What about planes that didn’t return—the planes that were shot out of the sky?

Wald recommended reinforcing the areas of aircraft that had no bullet holes. Because if planes were hit in those areas—the engines and the cockpit-- they didn’t return.

I wondered, and this is, I admit, more than a little inchoate, if there are parallels in the advertising business.

I noticed this first many decades ago when I worked at an agency (I’m changing its name to protect the innocent) let's call it Foot Cut and Bleeding. 
As always, I was a doer. I like to write things rather than talk about writing things.

I was in my mid-30s at the time, but I reckon I produced more TV and print than the rest of the agency combined. I’ve always liked working on a four-burner stove and keeping four pots boiling.

After about one year and having shot 25 commercials and 25 print ads, I got called into the boss’ office. (A CCO who added nothing to the creative process or product.) She played a commercial I had done on her monitor.

“Do you think that’s great?” she accused.

I was as realistic about work then as I am today. Not very much is “free swim” in our business. It’s not unusual if you work on big bureaucratic brands at big bureaucratic agencies to go through Rich Siegel’s metaphoric 17 rounds. And about 170 things you have to say. I often say, most ads are negotiations not communications.

Nevertheless, this one wasn't bad. I said, “I think it’s good. Smart, strategic and well-executed.”

“We’re looking for great here,” she said.

I probably said something snide and inflammatory but I don’t remember what.

Just a month or so afterwards, I was fired.

I realized back then what was reinforced yesterday with the bomber diagram above. It’s much easier to fix a problem if you don’t analyze the problem that needs fixing. It's much easier to fix something that isn't broken and break something that doesn't need fixing.

It's much easier to blame a creative team or a department if an agency isn't winning pitches than it is to look at what might be the real problem.

So Survivor-Ship Bias in this case says creative is the problem. That's what got hit by the bullets. That's what needs rejiggering, retooling, reinforcing.

That's what the data says.

But it depends on what data you're looking at.

So much depends upon it. 

Like careers, lives and morale.

No comments: