One day, back when my daughter was four, we told her that she was going to the dentist, and she got so excited. It seemed an odd reaction, but we never questioned it. After all, considerably less sometimes brings excitement to kids that age. As we walked out of dentist’s office following the appointment, she brightly asked, “OK, so where is the birthday party?”
We were a bit puzzled at the question, telling her that there wasn’t any birthday party, which disappointed her, as you might expect, and then we set to wondering why she made that connection.
It was only later that we realized that six months before, when she had her very first dental appointment, that she immediately went from the dentist’s office to a birthday party. So, her toddler mind had connected those two events as tightly as any other cause-and-effect she had observed. She had decided that all dental appointments conclude with birthday parties (which might actually be a better incentive to see the dentist, now that I think of it).
Why do I bring this up? Because it shows how human beings strive so hard to make sense of the world—we’re wired for it, which is how a four-year-old can do it so naturally. Even when faced with complete coincidences, we humans try to connect things into a sensible pattern. Unfortunately, a lot of events are random—noise, essentially—and don’t fit into any pattern.
And Internet marketing is no different. Without applying some rigor to what we do, we’re likely to connect a lot of birthday parties to dental appointments. And we’ll keep repeating the equivalent of trips to the dentist in our marketing for quiite awhile—which can be rather painful.
To prevent us from drawing lots of dubious conclusions, we need to use scientific methods. We need testing. We need to separate the data from the noise.
But how do you do that?
- Decide on your metrics. Your favorite might be return on investment, but if it’s profit or revenue instead, fine. Just pick your success factor ahead of time and rigorously measure it after every marketing decision you make.
- Let your metrics be your guide. This, for many, is the hard part. When you do something really cool and the metrics tank, stop. When that campaign you just love fails, kill it. When the dumb idea the intern came up with takes off, do more of it. If you regularly feel embarrassed and stupid about what the numbers tell you, then you’re doing it right.
- Super-charge your metrics. Don’t settle for “change then measure” testing. Use A/B testing to pit two alternatives against each other to choose the best. Or, better, use multivariate testing to try out dozens, hundreds, or thousands of alternatives to pick the very best. The testing is easy—the hard part is coming up with all those ideas. But think about how much faster you’ll find the best one.
Metrics aren’t perfect—they can certainly lead you astray sometimes, when you rely on small sample sizes or you just get plain bad luck. But more often than not you’ll be led in the right direction, which our feeble human brains can’t compete with. Save your brain for analyzing what the metrics say rather than doing correlation analysis. Then, it will be time to throw a party.
