Easily figure out what works: experiment!

A guest article by Jean Le Clerc for Experimentoran app developed for AB testing on Shopify.

If I told you to remove your navigation bar to increase conversion by 100%, would you do it? Probably not. But it's been very successful for some people.

yuppiechef saw a 100% increase in conversions by hiding their site navigation.
(source: VWO blog)

Experimentation can reveal surprising results - and not just of the kind that give you cool trivia facts for your next night out. They can actually move the needle on your sales.

I don't have time...


I know what you think: "Oh great! One more thing to put on my already overflowing 'To Do' list! There goes that hour I had left for sleep!" It doesn't have to be like this. And the return on your time investment is surprisingly high.

Running an experiment, is like hiring your very own pollster. Most times, for free. Let it sink in. An extra employee. For free. And she will never call in sick, or have a bad day. She will always perform at full capacity.

We're not talking polling like they do for elections either, where they only take a small sample of a bigger population. This pollster is thorough: it'll put all your visitors in the poll. Every single one.

You set it up, and then you let it run. You shouldn't run multiple experiments in parallel on the same store.  So really, you invest a bit of time for the set-up once every 2 or 3 or 4 weeks. That's it. And there's a very, very good chance a winner will be picked. Meaning something will be improved on your site.

I don't know what to test…


I can hear you think again (you really shouldn't think that loud...) "Ok bub: sure it's not too much time to set-up the experiment. But what do I test? How and when do I come up with test ideas, huh?"

I have a small trick for you. You have a technical or customer service line, right? (if you don't, maybe it's the first thing you should experiment with.) Use your clients as inspiration then!

This guy tells you he had a hard time finding your customer support number: is he the only person who can't be bothered to scroll all the way down to the bottom of the page? Or are all your visitors like him? A/B test the placement of your customer support line.  

That woman finds your newsletter signup pop-up invasive: if she's like most of your visitors, that's not good! Test it out and confirm. There are an unlimited number of things you can test.

Be curious. Have fun.

Don't be afraid to experiment - it's fun and easy and enlightening at the very least. There's very little chance you will break anything. A lot of experimenting algorithms actually prevent this from happening.

Google Experiments, for example, uses an algorithm that will start pushing a variation a bit more often, as soon as it starts to smell like a winning one. This means if one of your variation is a real dud, it'll be shown a bit less often. And if (I've never really seen this happen) things go very, very wrong, you can always stop an experiment with a single click.

You've probably heard about the benefits of experimenting (testing, A/B testing, ...) on your store, but deep down we all fear we will mess things up. The thing is, we're in a world of customization, and micro-improvements. Automatic experiments let you easily try things out and make micro changes to your store. Those add-up and before you know it, your bottom line will also improve, and/or customer satisfaction.

So go ahead. I know your time's limited. But I guarantee experiments are fun once you let your curiosity free. Test, prod, poke, learn!

Leave a comment

Please note, comments must be approved before they are published