Tuesday, May 19, 2015

Supposing is good, but finding out is better. (Samuel Clemens)

My life is full of conjecture. Now, some of it is often quite intelligent, and very well informed. That said, it is still someone’s opinion.

How can we get things out of the realm of conjecture? Well, one sure way is to rely solely on the numbers. And that’s where web analytics comes in.

One thing that web analytics sometimes misses, though, is the story behind the numbers. Yes, the users do seem to be dropping like flies on page 3 of the signup process. Why there though? What is happening on that page that makes them want to bail? What’s the story behind that?

Interestingly, there are plenty of companies and teams out there who, though they might be totally numbers-based up to this point, have no problem whatsoever abandoning that approach and conjecturing to their hearts’ content as to what those numbers might mean. And that kind of conjecture is typically no more informed or insightful than straight conjecture without any numbers. In other words, if you don’t watch it, there’s a good chance you’ll get pet peeves, “focus groups of one,” HIPPOs (highest paid person’s opinion), and anecdotes – instead of insightful, informed opinion.

Now, there are also groups out there who supplement their analytics with survey data. They might, for example, mine the survey data to find anything that might point to possible problems on that page 3.

And you can get even closer by providing a “give feedback” link on that page or by throwing in an intercept survey. The latter simply asks the user why they’re bailing. If it’s not pitched like pleading, you should get some pretty decent  feedback from it.

Another method, on the other hand, is to do some usability testing. A usability test always gets at the why. Just go ahead and put your users in the same position as those who were giving you your original analytics, ask them to think aloud, and see what you get. I highly recommend it.

Mark Twain doing some early man-machine studies

No comments:

Post a Comment