Let’s start today thinking about a language you don’t speak. It’s very possible, assuming that it’s written using the Latin alphabet, that you could pick up a book and begin to read out loud in that strange tongue. Of course, you’d have absolutely no idea what you’re reading. You can say the words on the page but you can’t explain what they mean.
Keep that image in mind as we change the topic to data. I can’t tell you how many times I’ve sat with clients and gone through their analytics reports with them and the aforementioned image has popped into my head. I don’t mean that to be derogatory to the people who pay me, nor does it mean that I’m fluent in analytics and they’re not. It raises a business point that is something we all need to keep in the back of our minds as data becomes more integral with everything we do.
Here is a small example. Most of us see “direct” traffic in our analytics reports. In theory, those visitors typed in the site URL or clicked on a bookmark they set on a previous visit. That’s a partial truth. The reality is whenever a referrer is not passed, the traffic is treated as direct traffic by Google. Think that’s an unimportant bit of information? How about in the context of mobile traffic not passing referrers at all (and I bet mobile is a big and growing part of your site traffic)? The point is that it requires both the knowledge that the “direct” bucket isn’t an absolute as well as some further analysis to figure out the truth.
I’ve seen the same sort of issues crop up in attribution modeling (what source was responsible for the sale). The groundwork for proper attribution hasn’t been laid and so the reports aren’t accurate. Sure, any analyst can puke out the data in front of them but the good ones – the ones who can interpret the words and not just say them – will tell you why there is a problem and fix all the links you’re putting out there to accurately reflect what’s going on.
“Keith,” you say, “I’m not a data scientist.” Neither am I. What I can do – and you probably can too – is to ask questions, especially when someone tells you they are dead certain about what the data is saying. Be sure they’re not just reading aloud in a language they don’t understand. Get beyond reporting and into meaning. It changes everything. Agreed?