What Analytics Can’t Tell You

There are lots of things that your website’s analytics can’t tell you, especially when it comes to human thoughts. The client who told me yesterday, “Here’s something that analytics can’t tell you” was referring to the buying behavior of people in his industry.  Analytics can tell us quite a bit, but there will always be information we can get better directly from human beings.

We can also get lots of misinformation from human beings. For one thing, we human beings are notoriously bad at remembering how we felt and what we thought. We’re easily swayed by emotions. And often we have no idea why we do what we do. We also all believe that we are special, and most of us believe that we are completely immune to marketing and are never swayed by merchandising, colors, or design.

So how can you make sure that the information you get from humans is reliable?

  • Make sure you’re asking experts. When I ask animal scientists what they’d do if they needed some goat gamma globulin, they can tell me. Accountants probably can’t. However confidently people tell you that golfers do this or teenagers do that, you’re better off asking a golfer or a teenager. And you’re pretty much always better off not asking your relatives or designers, because they just know too much about what you’re doing to qualify as what we call “naive subjects.” Naive experts, that’s what you want.
  • Don’t sway your results. When I asked animal scientists what they’d do if they needed goat gamma globulin, I said, “What would you do if you needed goat gamma globulin?” not “Would you type ‘goat gamma globulin’ if you were googling sources for blood fractions?” It’s okay to set up a situation–for example, when the scientists said, “I’d go down to the barn,” it was okay for me to ask, “Well, what if you wanted yak gamma globulin?” It’s not okay to offer answers to choose from.
  • Ask enough people. We may have a lot of things in common, we humans, but we still have individual preferences. If the first person you ask invariably reads the entire About Us section of all websites carefully, or thinks that he does, then he’s not typical. You’d find that out if you asked a few more people. If you only ask that one person, though, you’ll never know that he was unusual.
  • Compare your results with your analytics. If people tell you one thing and your analytics tell you another, then there’s something wrong somewhere. Your people are wrong, your analytics are wrong, or something else is going on that you haven’t considered yet. I’ve seen several cases in which surprising data was received because analytics were installed incorrectly. Comparing the answers you get from people with your analytics data can also help you avoid overgeneralizing from a few people who feel strongly about something.

More posts on this subject:





Leave a Reply