Michael has dedicated most of his waking existence to optimizing websites and decision-making processes. His hands-on approach and enthusiasm have made him a popular international speaker on the topic of Conversion Optimization. When he’s not preaching the CRO gospel, Michael spends his time helping clients improve conversion rates out of his office in wonderful Copenhagen, Denmark.
1. So Michael, who are you and what do you do for those or might not know?
Aloha – I’m the Senior Conversion Optimization Consultant at ContentVerve.com and Atcore ApS. I spend most of my waking existence optimizing and testing websites in order to better understand what really works in online marketing and business optimization.
When I’m not doing hands-on work for my clients, I travel all over world preaching the CRO Gospel at different online marketing-events.
2. Let’s dive right in there: A/B split testing, what mistakes to people tend to make?
Well basically I see people make a lot of the same mistakes that I used to make.
A very typical one is to test whatever feels good or seems like a fun idea instead of actually digging into the data and really understanding how people interact with your website and what’s keeping them from converting. There’s no law against running tests on a whim, but in the long run doing your homework and building solid test hypotheses is going to get you better results and make you more money.
Another very typical mistake is to call tests as soon as they hit the magic 95% confidence level. Confidence is just one out of many factors that go into determining the validity of a test results. Factors like sample size, number of conversions, standard deviation and the test period itself have major impact and must be considered. If you conclude tests based solely on the 95% mark, you are probably doing more harm that good in the sense that you are basing important business decisions on bad data.
A third very common mistake is to run tests without integrating the test data with your analytics setup. Most test tools just show you averages, and if you only look at the averages it’s safe to say that you are not getting the full picture. You really need to understand how your test variants are performing on different channels and devices, etc. in order to get the full insight and value.
3. What tools would you say are essential to any A/B split test?
If we are talking absolute essentials, there are really just two things you need, analytics and a split testing tool.
Web analytics is your bread and butter when it comes to anything related to website optimization. So I’d say a proper analytics setup is essential to properly diagnosing your website and finding out where your biggest problem areas are. Moreover, your analytics setup will help you track your improvements over time. And – as I mentioned earlier – you can only get the full picture if you integrate your test data with your analytics setup.
Secondly you need a split test tool. I prefer and recommend Optimizely.com. But there are lots of tools to choose from. Basically you need a tool that lets you set up tests and that helps you get reliable numbers so you can determine when you have sufficient data to call the test.
But apart from these essential tools, I use heat/click/attention/scroll maps, session recording software, eye tracking, user testing and surveys on a daily basis. I also use the phone a lot to interview customers and reps from sales and customer service.
4. Are there particular elements marketers should begin with split testing?
If you are completely new to testing, I recommend starting out with something reasonably simple that doesn’t entail a lot of code or design changes.
Elements like headlines, buttons and images are easy to run tests on and can have huge impact on conversion. Also site-wide elements like newsletter opt-ins are pretty easy and fun to test. Site-wide tests are great for low traffic websites because you can run the test across multiple pages and expose your test variant to a larger sample.
Another good tip is to go through your analytics data and look for pages that get a lot of traffic but have a high bounce rate. High traffic volume means better test data and shorter test periods. High bounce rate means that something on the page is making people leave the website instead of converting. Pages like these usually feature low hanging fruits that are well worth looking into. I also recommend setting up funnels in analytics so you can get an idea of where you are bleeding money.
5. You’ve worked on CRO for both B2B and B2C businesses, do you have the same approach?
Pretty much. Even though the target audience is different and the products are going to involve a different decision-making process, the process of gathering data, turning into insight, building hypotheses based on that insight, and then testing your hypotheses in the real world is the same whether you are optimizing B2B or B2C.
6. How do you see things in the CRO space changing over the next year or two?
The way I see it, CRO is the new “magic bullet” in online marketing. You can compare it to the state of SEO 7 years ago. Everyone is talking about it, everybody wants it, and all the sudden every agency out there has at least one “CRO Expert”.
Over the next couple of years I think we’ll see a lot of businesses spending a lot of money doing it wrong, and we’ll see a lot of CRO service providers rise and fall.
Hopefully in 2 years time there will be more of a conversion mindset where it’s natural to think in terms of conversion as a fundamental aspect of online marketing – instead of CRO being something you do as damage control when everything is falling apart.
On a personal note, I’m super excited to see how the market will develop. I’m constantly learning new stuff and hopefully the next 2 years will teach me as much as the last 2 did.
7. Any final words of advice?
Never forget that your target audience on the other side of the screen is made of flesh and blood. Ultimately, human decisions and actions determine whether conversion rates go up or down. If you take the human aspect out of online marketing, your optimization efforts will never reach their full potential.
As my all-time favorite online marketer, Dr. Flint McLaughlin, says:
“People don’t buy from websites, people buy from people.”