While the idea of changing the color of a call-to-action from blue to orange, or adding an image in order to skyrocket conversions by 150% would be great, in reality, you and I know that that almost never really happens. To get the most out of A/B testing you need to arm yourself with the right data. With dozens of variables involved in the optimization of a single webpage, without the right data, finding on-page optimization issues is like looking for a needle in a haystack.
Convert Optimization Stack
How do you find relevant data to plug potential on-page optimization bottlenecks?
Answer: Convert Optimization Stack.
The Convert Optimization Stack is a set of unique tools that retrieve useful insights on how a webpage is performing. The objective of the Convert Optimization Stack is simple: to provide data and measurements on various elements of a webpage to create a hypothesis for testing. The hypothesis is then tested using Convert Experiments (A/B testing software) to reveal if they are true. To visualize the concept, here’s a flowchart of how the testing process should look like:
Let’s assume I have a landing page that’s not converting as well as it should be, but I don’t know why. I start by using the Convert Optimization Stack to gather data on my landing page. Once the data has been gathered, I then create a number of hypotheses based on the output data from the Stack. Using Convert Experiments I test my hypotheses and if they were correct, I plug the friction points on my landing page and have the option to test further.
For example, let’s say I found that the lack of images on my landing page hindered the amount of opt-ins I received. By adding a single image opt-ins increased by 9%. I can further test by adding 2, 3, 4 or even 5 images to find the optimal amount to give me the best conversion rate possible.
If the Hypothesis failed or showed negligible change in testing, I create more hypotheses based on the data and test them again using Convert Experiments. Combing the Stack with the methodology above is the best approach to conversion rate optimization today.
What’s in the Convert Optimization Stack?
Landing Page Tools
Before you can start using any tools from the Convert Optimization Stack, you’re going to need a page to test. If you or your clients don’t have product or landing pages to test, to get the most value from the Stack you will want to use a landing page tool. Luckily, there are a number of great landing page products that work seemly with Convert Experiment. They include:
Test them out and pick the one that is best suited to your needs. Convert Experiments also works on a number of other platforms including WordPress, Magento, Joomla and almost every other platform you can think of.
Crazy Egg is a useful tool to help you understand how visitors interact with your website. The tool reveals which areas receive the most attention, how far visitors scroll down each page and identifies possible friction elements on a page.
Harry Fay Jewellery Boutique found that they had a very high bounce rate but couldn’t figure out why. After running Crazy Egg, they found a number of issues such as website buttons redirecting users back to Google, and social media icons that sent prospects away from their site instead of opening in a new window.
Using Crazy Egg’s scroll maps they found that their video content wasn’t as effective in keeping prospects engaged over product listings:
From gathering data on their homepage, they were able to make the necessary changes to increase user engagement and reduce bounce rate. Crazy Egg is an out of the box tool that can identify pain points and bad website practices that would otherwise go unnoticed.
Google Analytics is a free tool that tracks the footprint of every website visitor. It will show you how they landed on your website, how much time they spend on each page, the path they travel and a dozen more useful metrics.
One cool feature of Google Analytics is the ability to measure which web browsers visitors use to access your site. Here’s the data of one website I manage which receives a lot of mobile and desktop traffic:
The average bounce rate for mobile traffic using Android, Safari and Opera is 77%, with the average mobile user spending only 95 seconds on my site. My desktop bounce rate is 69.3% with the average visitor spending 161 seconds on my site.
It’s quite clear my website has mobile optimization issues which need to be addressed as mobile traffic is abandoning at a much faster rate. Another useful feature is their Visitor Flow map. This reveals data on how visitors landed on your website and the path they took before leaving:
Your website may have hundreds of pages and it can be difficult to know where to start testing. Look At your website’s Visitor Flow map to find pages that receive high drop-off rates and start there.
Website Speed Test Tools
Retail giant Amazon found that they could potentially lose $1.6 billion (yes billion) in sales for every second it took a page on their website to load. Walmart also learned the consequences of poor load times, and saw a 2% increase in conversion rate for every second they improved the speed of their website.
Not only does an average load time of 4 second see 25% of website traffic disappear, but it also impacts your search rankings:
Pro-tip: Don’t just check the speed of your homepage. Each webpage will consist of different elements and load times will not be static throughout. For websites with thousands of pages, I advise testers to start by running checks on their most visited pages.
In a world of tweets and emails, 75% of consumers say that calling a business is still the quickest way to get a response. There are currently two ways you track whether a prospect called your business as a result of your website or not. They are:
- To ask them
- Use CallTrack
While the first option may seem simple enough, asking the prospect what traffic source or keyword they typed to land on your website, or the webpage that inspired them to call is not only inappropriate, but could potentially put the prospect off from buying.
As the name suggests, CallTrack captures data on how prospects who called your business found your website. This gives you a better understanding of which keywords, traffic sources and landing pages are producing you the most business.
Anybody who picks up the phone and calls your business has a serious intent to make some kind of action on your website. This type of data is extremely valuable but often ignored.
Whether you or your client has an eCommerce store hosted on Magento, for reasons already mentioned in this article you’d want it to run as fast as possible. Here’s some of their best extensions to make that happen:
Full Page Cache – Considered the #1 extension to speed up a website’s load time. Using Full Page Cache on average makes a website 6-25x faster:
Image Optimization – High quality images can eat up bandwidth and dramatically slow down a website’s load time, and eCommerce sites are full of product pictures and thumbnails. Image Optimization for Magento reduces the file size of images without comprising quality and is compatible with most content delivery networks.
Sphinx Search – Never underestimate the power of a search bar. Screen Pages found that while less than 20% of visitors were using eCommerce search bars, prospects that did converted better and contributed to approximately 14% of sales revenue:
For more ways to increase the load times of a Magento website, read Guido Jansen’s awesome 101 ways to speed up your Magento ecommerce website blog post.
Once you’ve analyzed the data from each of the tools, in an ideal world you would find zero faults and rest happily knowing your website is 100% optimized. But as testers, we know that’s never the case.
Perhaps you used CrazyEgg and found that people were not interacting with your call-to-action because other buttons were in the way, or because the button wasn’t compelling enough. Using Convert Experiments you can carry out a number of tests to solve the problem to find the root cause and best possible solution.
Maybe Google Analytics revealed a particular webpage made 95% of prospects drop-off after landing. Testing that page using a speed tool revealed certain elements took several seconds to load and the users who waited for it to load, never interacted with them. Using this data you may want to remove that element or move it somewhere else to gain more exposure. Once again this can be tested and optimized through Convert Experiments. Once the problems have been found and the hypothesis created, Convert Experiments is used to run a number of tests to find the best solution.
To squeeze every last conversion percentage from any webpage, you must first identify the problem(s). Don’t think you can find problems and pain points by simply navigating a website through a web browser The foundation of any successful conversion rate optimization strategy starts by conducting research and gathering intel.
Once the problems have been identified, then Convert Experiments should be unleashed to run its magic. For example, if you found that a certain landing page is receiving an extremely high bounce rate but can’t understand why, run A/B tests that involve:
- Changing only the design
- Changing only the text
Maybe you find that the change in design showed a negligible change in conversions, but modifying the text saw a 15% drop in bounce rate. In this instance you found the message was the problem. Now you could further test short and long copies and even use videos to further reduce bounce rate and optimize your page. Just like a doctor diagnoses a patient before giving them the necessary medication, diagnose the issues on your website before you carry out the appropriate steps to remedy them.
Let me know your thoughts and feel free to share your tests results.
- 10 Aug, 2015
- Posted by Harvie Singh
- 0 Comments