One of the unsung heroes of World War 2 was a gentleman named Abraham Wald. You wouldn’t know it, but he’s responsible for saving thousands of American airmen lives because of data he didn’t have.
As the allies made gains in Europe they had the ability to add armor to planes for resilience against anti-aircraft flak. But where to put it? And how does this apply to modern day digital marketing?
They cataloged and mapped damage of all the planes returning to see where they were hit more often. That’s when Mr. Wald said STOP! We only have data on the planes that returned and no data on the planes shot down. So Mr. Wald looked for holes in data based on the surviving aircraft because given a random distribution, the gaps on the return planes were likely sensitive spots that caused their downing of planes.
Mr. Wald knew he was dealing with a survivor’s bias in his data because he was looking at an incomplete set: Only the planes that could return. By subtracting the known data from an expected random distribution they knew exactly where to improve armor and increase plane survivability by 30%. That’s a lot of hardware that doesn’t need to be rebuilt, and—more importantly—a lot of air crews saved to fly another day.
How does this help you, a 21st century business owner or marketing manager? Well, you need to understand your data is also incomplete, and any outcomes you draw from success are only based on observed success. You have to dive deeper into the changes in non-successes to see where you can really impact change. For example:
You run a campaign, generate more sales, but your site’s conversion rate declined. Some would say the campaign worked because of more conversions, but if the conversion rate declined your campaign may have reached the wrong audience. Even worse if you targeted the right audience and the conversion rate declined. The increased sales may have been reminding certain people to take an action with you they were going to take anyways, but the overall messaging turned away other members of your audience.
That’s where D2 comes in. We channel our inner Abraham Wald and look to see how people interact with your message to see what about the message worked or didn’t work. It’s not surface-level, but once you see the hidden spots you can’t unsee them again, and you’ll run your campaigns in an entirely different way.
Book a time to talk to our lead analytics guru who can walk you through what he sees in your data and provide suggestions with how to spot data bias. The worst-case scenario is you learn something new about analytics and can make better decisions. The best case scenario is we help you uncover a treasure trove of information to help guide you to the next major campaign success story and generate a bunch of new sales.