A few lessons ago, I said that the first type of analysis you can do in Google Analytics is to check on the performance of your site: to observe.

Are you getting enough visits from your Facebook ads?
Are your new item descriptions resonating with men?
Has your bounce rate decreased after you started cross-promoting your listings?

If you know what “success” means for you, it’s not hard to set up a report to look for it.

What’s tricky is knowing whether that number of visits or engagement rate or bounce rate actually means what you think it means.

Yep, because you can have the BEST engagement rate in the world and it can still mean diddly squat!

Sometimes things just happen

When you’re looking at your numbers, it’s easy to get caught up in the idea that everything you see is happening because of something to do with your shop.

The high bounce rate is because of your banner or the low engagement rate is because you’re attracting the wrong visitors.

But you know what has more impact than any of those things?

Real life.

Any one person is more likely to get dragged away from the computer by their children or a phone call. And occasionally, the Facebook algorithm will show your ads to people who are not your ideal customers despite your best efforts at defining a perfect audience.

This is a basic way of talking about statistical significance. That sounds like something terrible and math-y but it’s pretty simple.

Say we compare the Bounce Rate between visitors from Canada and Italy, and discover that Canadians had a 75% bounce rate but Italians just 65%.

Statistical significance is the probability that the difference between the two Bounce Rates is actually due to the visitor’s country, and not just because the Canadians had noisy kids or slow internet.

Ultimately, we’re trying to work out whether the difference is real and will continue over timeIs it worth making a business decision about?

Oh no… how do I apply this without doing maths!?

There’s a lot of maths and scary equations you can apply to work out statistical significance, but as a small business owner, you have better things to do. Just keep a few things in mind:

  • If you’re looking at any numbers under a few hundred (thousands are even better), it’s more likely that what you’re seeing is due to chance (real life).
  • Do not make high impact or costly decisions based on numbers under a few thousand. (eg. Doubling ad spend or removing half your listings.)
  • Watch numbers over time. Make small tweaks and watch, watch, watch. Let the numbers build up.
  • Feel free to assume and guess, if it won’t cost you much. The best way to work out if an effect is real is to “test” it by making changes. (But make only one change at a time!)

It all makes sense when you think about it. A 50% bounce rate on only two visits just doesn’t tell you anything. (In fact, you probably figured that out already!)