Data Always Lies

data always liesOne of my favorite micro roles in a job over the last few years, was at a San Diego SAS startup that focused on a very technical application of data in a high impact business marketing. Amongst my many, many roles and responsibilities I was able to spend a great deal of time with our Data Scientist. He was a breathtakingly brilliant double PhD brainiac. In my role of Marketing I was excited with the opportunity to have access to a window into the massive amount of data, and better yet a clear and impartial interpretation of the data.

We collected a mind-bindingly large amount of data. Crazy amounts of data. As a marketing wonk I was in heaven. We love data. Especially data on our customers actions, wants and needs. Anonymized data is even better for us marketing geeks because it tells us big stories, not tiny stories. It was fascinating to think hard about what our customers wanted to do, what challenges they faced and where they could make improvements and then think “I wonder what the data says?”

So with a resident Data Scientist sitting next to me, with access to massive mountains of anonymized data, I could turn and ask him. He could do some crazy complex algorithms and data dances and the data would spit out an answer. Awesome! The problems arose when the CEO would ask for what the data showed us about subject A or problem Y. He would task us with finding out what the data said.

And invariably the data would spit out an “answer” that was not even close to what we thought or wanted.
matrix

Fast forward to the conclusion because it is late and I am tired.

Data always lies. It always tells you the answer to the question you ask it. That sounds simple. it’s not. If you take a giant pile of data and then ask it to tell you the answer, it will tell you nothing. Ask it a specific question and data will tell you very specific things. You may not like the answer, but it will tell you the answer.

However data only tells you the answer to the questions you ask it. In other words you can change the question and the data will give you a different answer. You can ask myriad different tweaks to your question and you will get myriad answers from the exact same data.

That;s because data always lies. It depends on the question you ask it. What marketing wonks is not more data, we need better questions. The data only answers the question you ask it. It doesn’t volunteer answers to unasked questions.

data always lies

Data only answers the question you ask it. So polish up your question asking brain and work 10x harder on the questions than you do on the data.

Thoughts and Counter Thoughts?