A big part of successful optimization comes from the tools used to run the testing…
[Interview]: Easy Testing Tools and Personalization with Tim Ash
Tim Ash, Landing Page Optimization host, interviews Dennis van der Heijden, founder and CEO of Convert.com.
Living in San Diego, Tim wanted to know how Convert.com got started, and of course, why in Cancun.
“Well, I’m lucky to live there,” said Dennis. “We were a lead generation company that ran into some trouble in 2008 due to the crisis. We built some workplace plug-ins for conversion optimization and became a company. We got funding last year and brought in all of those involved, who were living in different parts of the world into a neutral agreeable place which was Cancun. There are people from Romania, the USA and Holland, and we are all in the office working hard and enjoying this beautiful spot. On Fridays we go to the beach, have a drink there, and not talk about work.”
Tim’s slightly jealous, and the two move on to discuss the 4 pillars of landing page optimization:
- The standard bag of tricks: copywriting, headlines, visual design, user experience, getting rid of stumbling blocks.
- Pending third party data that you know anonymously about your visitors.
- Changing what you know about the site based on your inner past interactions with them; and
- Behavioral targeting and seeing what they actually do in the site and then changing the site experience.
Putting all of these together, you have a truly one-to-one experience for people and having discovered the power of this, Tim asks Dennis to describe what makes Convert.com software unique and where does it fit into the big picture?
Dennis goes back to 2005 when they lost about 50% of the visitors, and the previous company was struggling. “We picked up something very basic , depending on what keywords people were using on the adwords campaigns and natural search, we got an idea of what they wanted, combined by which pages they visited, and what time they were in specific pages, we were able to manipulate and slightly change those pages to fit their needs, in other words, they left a “sent” behind.
“We had multiple domains in that one like CRM software boxes and ERP software boxes, whistled the lead, and if people went in looking for CRM and later in the week came in with an ERP question, we mixed those two blocks of information, and those visitors would get asked if they wanted the free ERP box that selected an ERP system that connected really well to CRM systems. After a couple of visits they would be ready to personalize the box with the information that we had, and got a good lift in conversion,” said Dennis.
A lot more people wanted that, so they built a product; Convert. Dennis describes it as the A/B testing tool for agencies and experts, combined with an advanced segmentation option. For example: If they want to run a test on a specific set of keywords used in the last year, plus the behavior on the site, and as soon as they discover something and they Google analytics to find out why the conversion rate of fb visitors is less than the conversion rates on the adwords or natural search, so they can actually run a test and everyone that came in from fb to their site, today and in the last year, if they ever come back, are going to get bucketed and they’ll be offered an additional set of plug-ins that relate to fb, to see if that can get a lift in the conversion rate for that particular source. It’s e- testing, but very targeted.
Tim is interested in some of the common things that can be done with segmentation in the package besides standard split-testing .
“Convert has a whole set of lead generation companies that do something very basic and logical. There’s a whole group of people looking to increase their list time through emails, so if they convince someone to leave their email address through a sign up box, Convert will suggest that the next time that visitor goes into the site, the sign-up box will be replaced with a newsletter box.“Loyal” visitors will then follow a lead-nurturing pattern that takes them into the next level, like downloading a white-paper to manage the engagement for the purchase of a training program, or a PDF and the like, which leads to the purchase of a program. The interesting thing here is that with Convert.com, the box continues to change accordingly, so it’s no longer a wasted space on a site.”
After a commercial break Tim Ash clarified what marketing automation tools were, for the benefit of those who weren’t familiar with the term. Basically it’s a tool that provides a 360 degree view of anything that someone has ever done or downloaded on a website, emails they’ve opened, and with that info gauge their level of interest, or where they are in the funnel, like at the awareness stage, or considering and comparing, or are they ready to act. In the case of Convert, Tim wanted to know how Convert mapped this onto what the site should actually be doing to know exactly where they are in the funnel.
“The easiest thing is using a specific URL structure and the email newsletter, which allows us to pick up where the email system leaves off. It basically means for a user app on any commerce site if their shopping cart is full, but they haven’t checked out, we can signal your marketing system, if it’s ready for that, that their shopping cart abandonment process should start, and as soon as they clicked on the site, we can actually say they went back to the site but still haven’t purchased, so that a new set of emails should be triggered, so even if the system is not integrated we can play really well those abandonment shopping cart and lead-nurturing campaigns.”
Tim further commented that it’s two-way signaling. Convert is signaled what stage the visitors are in, and Convert signals back new actions the visitors have taken or changes in the stage that the visitors are in, back to the marketing automation software.
Dennis agrees and adds “They can pass it using URL or we can pass it back using java script, but in core, Convert is an A/B testing tool but it has those behavioral rules and segmentation options that allow especially for marketers and people with more expertise in A/B testing who are looking for a little bit more than the basic A/B testing tools out there, and simpler than the very large enterprise tools.”
Tim comments about simplicity because doing large scale implementation involves things like test and target, and interwoven autonomy optimist and platforms. Basically you are talking about a months-long integration, very expensive, software licensing, not to mention a bunch of customization and professional services on the back end, so it’s not something one starts doing, so it would be interesting for the listeners to hear Dennis tell about the difference in approach in the setup and ease of use offered by Convert .
“Convert has reached a level of A/B in which all you need do is to paste a java script in the header of your website to get started and that’s all. The rest are things on our end, and we try to stay away where you don’t have to implement anything else but that java script. A/B testing is becoming something that everyone can do, there is no excuse anymore, there are very simple tools, and as soon as you pass your first test, we usually get a lot of people who just want more, and when you see the options out there and you had some external systems out there that you want to segment on, or signal then these tools should be very simple. Just paste the java script up there and for the rest its implementation time load of five minutes.”
And how does Convert hook into third party systems to get those additional business rules or triggers for swapping out content in the first place?
“We try to build integration as simple as possible, for example what we got right now is integration with KISSmetrics so we signal KISSmetrics all the day out, and for example pick up items from Google analytics without you having to install any additional codes, and Google will pick up any revenue track analytics and will display all the revenue results inside testing, so there’s a lot of things you can do nowadays without adding any additional scripts to the client site. Listening to other people’s java scripts allows you to integrate really well, so we can pick up goals in Google analytics, we’ll pick up revenue from Google analytics, order items, all those items you can pick up and signal back so you can signal back custom variables to Google analytics for people to slice and dice the A/B testing data we provide to one another. Most of these tools have very powerful capabilities, so it’s just a java script away and from that moment on these integrations are usually done with a checkbox and then it’s integrated.”
On the topic of behavioral targeting, Tim wants to know if Convert can change the site experience for someone based just on what their actions on the site are, and what they’re doing in their current session. Has Convert or their clients ever tested something along those lines?
“We’ve got people that based on the behavior they can monitor clicks on a specific event and either redirect them or change the experience based on that, so if you know a specific person hangs out in a category for 10 page visits right now, we are able to test offering them a free shipping coupon. Just putting in this extra banner in there we see they are visiting products more than $100, it is worth for us to offer this coupon to this person. We are not a personalization tool but we can test if these offers move the needle. So it’s an A/B testing tool but it has behavioral rules, and that’s what we’re testing and our clients are testing.”
Convert basically proves advanced segmentation, if it works then the client is ready for a personalization tool.
“We start where other A/B testing tools stop. People come to convert looking to see if they are ready for the next level. Once they prove they are ready, if personalization makes sense, then, personalize.”
After a second commercial break and more on the lighter side of things, Tim asked about Dennis’ globetrotting ways, which have taken him to over 60 countries. His first experience was traveling from USA to Japan and China, and in his words “It was a bit of culture shock, given that Japan is the most organized country in the world and China was the least structured one in the world at the time, but both are fascinating places.” The strangest experience was, “Having an elephant walking past my lodge each morning, which was kind of scary.” That was in Zimbabwe, very near Victoria Falls where this local elephant used to walk around the village looking for something. People told him not to panic, but it was quite a surprise to wake up with an elephant next to him.
Finally, Tim asked Dennis for some tips for people that are doing testing programs in terms of approach, pitfalls to avoid and practical advice on using testing tools like Conversion.
“If using a tool like ours, it’s not about testing. Conversion optimization is not about testing. That’s about 10% of the entire process. Looking at what our clients do and the most successful ones we have, the 90% of their work is getting the problem right. Our pricing is more expensive because we keep the button testers away.
It’s sexy to publish cases and it is sexy to show 150% increase, and it’s super sexy to show that the President makes $60M, yet this sends out the wrong signal about testing tools, like it’s the way to only way to go forward. Dennis differs.
“Testing is a tool to prove that the hypothesis you have is right. So it’s usually that where people get stuck. Getting a problem, looking at your analytics, finding where people get stuck, talking to them, researching, doing eye-tracking, doing customer interviews, doing user-testing, and only at the very end you have a hypothesis on why things are going wrong, and you try to find a solution and prove it with testing. Don’t test something because you saw it.
Tim, being of the opinion that when coming up with solutions or ideas for what to put in your tests, champions radically different approaches to separate out the different versions, not to have slight variations on the theme or a timid approach if you are doing split testing, and asks if that is the way to go, to which Dennis agrees.
“Internal stats of people using our tool show 50% use A/B testing, wysiwyg editor, another 30% using split testing and only 20% multivariate testing, so we know 80% of the people are following your advice, saying do A/B testing, do radical changes, and as soon as you know that you are in the right direction fine tune a multivariate test and figure out which of these specific elements were the cause of their A/B test to win. I think this is the right approach, radically different tests based on a promo statement and hypothesis, and afterwards fine tune to see what particular message in that A/B test maybe moved the needle, and if there were some underperforming elements in the A/B test.”