Since I began working at Yoast, I’ve been busy with a lot of things, but my personal favorite definitely is this one: conversion. Maybe this is because of my background in behavioral science. Or maybe I just prefer bigger numbers to smaller ones. In any case, the fact remains: I’m hooked. Now, for the people thinking I’m here to convert them to any kind of religion, be at ease. Conversion, in this case, means converting visitors of your website to actual customers (or returning visitors). And optimizing the conversion rate is an ongoing process on our website. During this process, I’ve come across a few things I want to share.
Don’t believe everything you read
Maybe it’s because I’m looking for it, but I’m seeing more and more blogs, books and ‘experts’ in the field of conversion. This is fine, of course. In fact, I’ve learned a lot from all those people. But there’s something important you have to keep in mind at all times: what they’re saying or advising might’ve worked for them, but that doesn’t mean it will work for you.
So how will you know what works for you? Good question. The answer is actually simple: you should test that! Yes, this is a conscious choice of words, as it’s also the title of the book by Chris Goward. That book really changed my idea of conversion and made it understandable and clear. But most of all it convinced me of the need for testing. Use the ideas and advice from those blogs and books as hypotheses. And then test them.
The easiest way to test is the A/B test. A is your baseline, or in other words: what your website looks like before you change anything. B is your variation to A, so a changed button, changed text, or anything like that. You will then test these against each other to see which one works best. Now you can add as many variations as you want, making it an A/B/C/D/etc test. However, when using a lot of variations, you’re better off using multivariate testing. But that’s something I won’t go in to here.
At Yoast, we’re currently using Convert for our A/B tests, and we like it very much. It’s clean and simple enough for even me to understand. There are several others out there offering the same packages though. But using tools like these, you can actually set up variations to a page (or several of them) and see which variation converts best. And this immediately takes us to the next point.
Revenue trumps conversion
Now, no one can disagree on this one. If you’re making money with your website, the top thing you want is to make as much money as possible. This means revenue is more important than conversion. What I mean by this is the following: if having just 2 conversions a day is making you more money than having 4 (lower value) conversions a day, you’ll go for the 2 conversions.
The idea is simple: when testing hypotheses, the one making you the most money wins. So don’t just focus on the number of conversions, be sure to check what the tests are doing to your revenue as well. And also be sure your revenue is being calculated the right way. So never rely on just one tool, when setting up your tests. The best is to have an exact listing of all your transactions (conversions) with an amount of money attached to it. This way, you can always double check the revenue and be sure it’s being calculated the right way.
This may not exactly be true, but patience does play a very big role in conversion rate optimization. It’s never a good idea to rush your conversion rate optimization, as you might pick the wrong variation or option. You should always leave a test running for at least a week, and preferably two, no matter how much traffic you get. The reason for this is simple: people’s online behavior is different on the weekends than it is during the week. I can prove this with a simple graph of the organic traffic of yoast.com:
Now guess which days are the weekends. Freaky, right? So your tests should run for at least a full week, to account for differences between the days. The longer you let a test run, the surer you can be of the results. And believe me, this requires patience. I’m from the “I want it and I want it now” age, and it can be hard to not just make the changes and hope for the best.
Make it a science
This last one I cannot stress enough. Don’t be afraid you’ll get in over your head; it’s actually really easy. What you have to do is formulate hypotheses for what you’re testing. And I understand it if you won’t dive into the scientific literature to find references and support for your claims (although this would be awesome). Just be sure you know what you’re expecting the outcome of the test will be and why. Redundant as this may seem, it’s actually really important.
Hypotheses are not things you just think of yourself. You have to have a reason why you believe something will work, other than your gut feeling. So if you really want to make it scientific, you could dig into the scientific literature. Believe me, there’s a lot of it out there that could be of use for your conversion optimization.
However, as I said, you don’t have to do this. In fact, there are quite a few people who probably already did the hard work for you. For instance, the amazing people at HelpScout have already done all the research in the strategies of pricing. So a good way to come up with hypotheses is to read blogs and books and see what they say worked for them (or their clients). Usually, there are good (psychological) explanations as to why this worked.
The next step is to formulate hypotheses from these sources. What will you be changing on your website, and what do you expect will change and why? Make sure you use your sources to explain the ‘why’ of these expected changes. This is what makes it a science: backing up your ideas and expectations with actual results from other people.
Formulating hypotheses makes sure you think about the tests you’re about to run. If you don’t put enough effort into thinking of logical hypotheses, you will lose precious time trying variations that make no sense.
So it’s really simple: you start with a good hypothesis, then you test whether you were right and then you implement the changes.
Now it’s up to you: have you already formulated some nice hypotheses, which you’ve already tested? If so, please do share!