Karl is a former rocket scientist with a PhD from Cambridge University, England. He co-founded Conversion Rate Experts to show businesses how they can profit by applying scientific techniques to web marketing. Conversion Rate Experts is now an international consultancy whose client portfolio includes Apple, Sony, Google and Facebook. We wanted to give our audience the chance to get to know him a little bit better in preparation for our webcast, so we asked him a few questions.
(Psst…the webinar is now available on demand. Hooray!)
How did you make the jump from being a rocket scientist to a conversion rate optimizer?
I feel that conversion rate optimization was made for me. It’s the point at which several of the themes of my career converged. While pursuing a science degree, I worked one summer for a local business and won a national prize for entrepreneurship for the work I did there. I then continued with university and did my PhD developing high-temperature materials for combustion chambers.
But about 10 years later, the guy who owned the business I had worked for in college asked to meet with me. He offered me a job running one of his companies, which I took. So in one jump I went from science to running a cell phone company. It was like a crash course MBA.
That’s where I met my co-founder, Ben Jesson. We tried every technique for growing the cell-phone business, and conversion rate optimization was the one that was the most effective. It allowed us to triple the company’s revenues in 12 months. Shortly after, we published an article containing 108 tips for conversion rate optimization. The article went viral, and the following day Google invited us to partner with them. And Conversion Rate Experts was born.
Is there anything you learned in science that applies to CRO?
Loads actually! At one point in my science career, my job was to troubleshoot large-scale industrial processing plants. That’s where I first used multivariate testing. My first multivariate test was in a plant in Paris. The cost of making a mistake on a production plant is huge, so everything has to be done right. Multivariate testing is useful for web marketing for exactly the same reasons.
At that same time, I became a Chartered Engineer. Process engineering is all about designing processes that are extremely efficient. At Conversion Rate Experts, we have discovered that many process engineering techniques are extremely applicable to web marketing. After all, a web marketing funnel is a process, with inputs and outputs. On a different level, the actual creation of a site is also a process (the stages for which include ideation, prioritization, wireframing, coding, QA, testing, etc.). Our methodology is all about introducing the most useful tools and techniques of engineering and science to optimize these processes—and ultimately revenue.
So, yes, my experience with science and engineering has been surprisingly relevant.
So we should be hiring out of NASA?
That’s not a bad idea. In fact, the head of our research department is an ex-aerospace engineer. We plucked him out of aerospace and he’s doing an amazing job with all the research.
Can you tell me about an engineering process that you’ve specifically applied to conversion rate optimization?
One thing that’s really interesting and useful is the theory of constraints, which is the science of looking for bottlenecks in a process and removing them. Picture an hourglass. If you were trying to speed up the time it took for the sand to go to the bottom then you would need to widen the bottleneck.
People understand that there are bottlenecks constraining the rate of their business’s growth, and that it’s important to work on the them, but what most people don’t understand is that in many cases the bottleneck is the only thing to work on. Working on a non-bottleneck often yields no increase in throughput whatsoever.
To some extent you can apply this to conversion funnels and CRO, but primarily this applies to the business side of your company. For instance, ask what is keeping you from doubling your revenue this year. You may well find one bottleneck that is preventing this from happening. Perhaps it’s the rate at which you can create web content. Perhaps it’s the time it takes to get approval from your “brand police.” Perhaps it’s the time that the IT department takes to push a new test live. Pouring more sand into the hourglass won’t make it go through any faster (it actually can make the problem worse). You need to expand the bottleneck. By doing so, you can grow your company surprisingly quickly (here are some examples of what we’ve achieved). We help our clients to increase their conversion rate, but some of them report that the biggest, longest-lasting benefits come from how we’ve helped them to implement more quickly and effectively.
The title of your webcast is, “Are Your Visitors Losing the Will to Live?” Why do people say things on their websites that they would never say in real life?
There’s a strange disconnect between what people would say to a human and what people actually write on their website. The most likely reason for this problem is that people can’t actually imagine speaking to the large number of people that will visit their website, so they resort to generic, even robotic-type language.
An exercise that we often carry out with website owners is to tell them to imagine their service is in a physical location and that a new prospect just walked through the door. The prospect has questions about the service and would like answers. We tell the website owner that they must answer the prospect using only what’s written on their landing page. Usually within a couple of words they are so embarrassed that they won’t even do it. They would never say those words to a real person.
So how can we help people write for human beings?
As Tindersticks sang, “Stop pretending, stop pretending, stop pretending.” Since it’s much easier and natural to speak to one person, then do just that.
For example, when one of our consultants wants to write a broadcast email, they pick a specific person who matches their target demographic, then they write the email to that one person, using their normal email client. And then they press “Send.” This may not sound sophisticated, but it’s really quite elegant. Because if you sit down to write an email to 10,000 people it immediately becomes artificial, and you don’t harness the part of your brain that allows you to naturally communicate. (“Cognitive scope limitation” is a term that describes how human minds can’t handle the concept of thousands of personal connections.)
It’s similar to acting. It’s almost impossible for an actor to act 100% realistically because what they’re doing isn’t real. In our case, we take the barrier away and stop talking to a generic group of people, and actually force ourselves to talk to a real person. You’ll become less like a marketer and more like a human. Which you’re probably already very good at.
Your organization does a lot of testing with multinational/multilingual websites. What challenges and solutions have you discovered while testing in different countries and in different languages?
Our consultants speak nine languages, and we have worked with clients in 15 different countries, so we have a lot of experience with this.
The main mistake that people make in internationalization is that they assume things will be entirely different in other languages and culture, when often there is not much difference. When you assume complexity from the outset, you begin with a huge mess. It’s surprising how much you can do things consistently from country to country.
One example is a client we worked for that had offices in both the US and Japan. The Japanese office told us that the “American” way of marketing just wouldn’t work for Japanese customers. We told them to trust us—or at least to trust the split-testing process—and we simply translated the English on our optimized landing page to Japanese and left everything else the same. We managed to double the Japanese conversions with a page that the Japan branch was adamant wouldn’t work.
Guess what: Star Wars, Harry Potter, and iPhones are popular in any country in the world. They have universal appeal. Ultimately, people are very similar even in different cultures, and good content is often universally liked. So the first lesson is “Don’t assume complexity unless you have evidence for it.”
Obviously, some localization is always needed. Using the Harry Potter example, this is analogous to making sure you do a good job of translating the book into the local language, and taking into account regional variations. First, ensure that you have gathered customer feedback from actual prospects in the local market. Also, user-test your content on real visitors. Here are a few examples of regional differences:
- Shipping is more complicated in Australia because of its huge size and small population density.
- Privacy restrictions are much greater in some countries, like Germany.
- Some phrases, when translated into German, do not easily fit within the allocated space (for example, “Sign up now” requires a larger button when translated into German: “Jetzt registrieren”).
- Some countries have novel payment processes (which is a challenge for our client daFlores, which ships mainly to South America).
Ultimately, you need someone on the team who understands how to balance the huge efficiencies of standardization with the practical opportunities of localization.
Why do some websites make people “lose the will to live?”
That phrase actually came from a user test we carried out. The participant really did say, “I’m losing the will to live.”
It’s a good reminder of how “normal people” (i.e. people who aren’t web marketers) struggle to use sites, and how emotional it can be for them. If you’re a web marketer, then most of the people you know probably use the web regularly. So you don’t often get to see normal people using the web. Things that might be obvious to an experienced user are incredibly difficult for normal people.
Marketers develop huge blind spots. One time I was interacting with someone who was explaining his site in a very sophisticated way. After looking through the site for a few minutes I told him that I had no idea what his company did. He was obviously quite surprised.
That’s why user testing is so helpful. After working on a site for even just a few days you can lose a fresh perspective. We call it the “house visitor” effect. When you have a visitor in your house, the minute they walk in the door you start noticing things that you didn’t notice before, like the pile of books in the corner or a mark on the wall. You start seeing everything through the visitor’s eyes.
One problem that often arises is that you design a page that requires scrolling and the user doesn’t realize it and leaves the page without seeing most of it. This is such a common problem, we’ve written an article on ways to make it evident to the user that the page is scrollable.
You’re known for making great toy recommendations. The Syma S107 Helicopter you recommended for the “children” has been quite a hit. Any new recommendations?
A website that has been great at keeping the “kids” (i.e., me) entertained is bombermine.com. It’s a massive multiplayer version of a retro arcade game.
Also, it’s not a toy, but we’ve been into standing desks lately (apparently, the latest research links sitting all day to obesity, diabetes, heart disease and cancer). We’ve bought a few Ergotron sit-stand desk mounts which allow you to adjust for when you want to sit or stand. They attach to pretty much any desk, and it’s surprising how liberating they are.
One other recommendation for baby toys is the Baby Mozart DVD. Many of us have recently had babies. This DVD kind of hypnotizes them for hours. Which can help to keep the noise down.