4 Warning Signs You Need To Carry Out A/B Testing

The following is a guest contributed post from Natalia Selby, the marketing manager at Mediahawk.

The grass isn’t always greener on the other side, but is it worth finding out if it is? This is certainly true of your website offering and one way to achieve this is through A/B testing.

An A/B  test, also sometimes referred to as a split test, is where you compare two different versions of something like a webpage, an email, ad copy or even smaller elements like a headline, to see which performs better. It’s achieved by using two different versions of the same content for the same period of time and comparing the results. A/B testing can be a simple and cost effective way to ensure you are maximizing conversions from the traffic you’re already getting to your website.

Performing an A/B test isn’t always conclusive but looking at an average of the results across 25 case studies it, the winning combination improved conversion rates by a staggering 48%.

So, if you’re still not convinced as to whether you should test different elements of your site, below are the four warning signs that you need to be aware of and carry out A/B testing

  1. Your website isn’t converting

Ultimately, conversions and enquiries are what your website is there for. It shouldn’t just be seen as the bait on the hook to entice traffic, it should be your net to land that prized catch. If your website isn’t optimised and working correctly a customer can’t or isn’t able to fulfil the required end goal.

Conversion rate is based upon the percentage of customers performing a required goal against the amount of traffic that your website receives. The required goal can be purchasing a product, requesting a service, data capture or anything you specify. In order to use an A/B test to increase this figure you should first look at your key service or sales pages. These are the pages that drive your conversions so it is critical they are working as they should. If your website is experiencing high traffic and your conversions rates are consistently low then testing these pages is a must. It’s important that key calls to action are visible and obvious, as well as removing any excess imagery and copy, or barriers that obstruct a seamless path to purchase.

Simply removing one image can increase conversion rates, for example, iCouponblog removed a rudimentary secured transaction image from their checkout page after finding a high bounce rate on that page. This resulted in a whopping 400% increase in conversions.

You can also test to see if you can improve other calls to action on your website at the same time such as social media share buttons, links to key pages and download buttons. Simplifying the “action” process is usually a logical area to test as these types of interactions are all part of the sales cycle and can result in a sale being made in the future. For example, you may have a button that allows potential customers to download a lead-generating PDF document containing technical specifications for your product. Having looked at the conversion rate for this button it shows a low percentage of traffic to that page haven’t followed through and clicked to download the PDF. You might think that this conversion rate could be improved by changing the design of the button or moving it to a new more prominent location so you could make these changes temporarily and use the amended page for a period of time to see if this positively affects the conversion rate.

  1. Your bounce rates are increasing month on month

Bounce rate is the name for the percentage of visitors to your website who navigate away after only viewing one page. Reasons for ’bouncing away’ from a site are varied and don’t always mean the worst, but sometimes it’s because they didn’t find what they are looking for on your website, or it’s not immediately clear what it is that you provide, or they don’t know where to find it. These are common reasons why a visitor navigates away, potentially never to return.

To establish whether this is an issue you are currently faced with, look at your website analytics to identify what the current average user journey is and compare it against the expected user journey and see if they add up. You should also check the average time on the website and compare it to someone who has completed the desired customer journey and, finally and more obviously, check to see when your bounce rate started to rise and check to see if this coincided with any changes or site updates.

If bounce rates are rising, this is usually content/usability related. The potential customer managed to find your website easily, but once they arrived on the landing page they weren’t engaged and didn’t or couldn’t find what they wanted, so they navigated away. You could consider splitting data via device, if bounce rates appear to be higher on mobile then you might want to invest in better mobile- optimised pages which can be achieved via an A/B test of mobile landing page designs.

  1. People are calling from your website asking the same questions over and over again

It might sound fairly obvious to say that if multiple people are calling you asking the same question, this could highlight that something isn’t obviously clear on your website. Sometimes though it can be a lot tougher to uncover issues that are more difficult to replicate without asking too much of your customer. For this reason there is something called call tracking software, which can be added to your website easily. The idea is that the call tracking software provides an individual telephone number for each unique visitor to your website and this allows you to track the user journey on site up until they call you. The software will show you data such as which channel the caller came from, the pages they visited and most importantly, which page they called you from. This can be crucial in discovering where they are experiencing problems. As well as this function, it keeps all customer correspondence in line with their user journey, making it easy to see if the people answering the customer queries on the telephone are successful in resolving them. It will also record calls for you to later analyse.

Once you are aware of the issues facing your website, you can perform an A/B test on two better optimised versions of a problem landing page and decide which is best suited to resolve the issues.

  1. Low traffic to category pages

Low traffic to category pages can be a sign of a poorly optimised user journey. A reason for this can be because the navigation is difficult to use or is not offered in a logical structure to send visitors to the desired landing pages. This can cause you to lose visits even before they get a chance to browse.

To solve this, consider amending the navigation of a variant to test. Try implementing a simpler navigation to pages lower down in the hierarchy, or potentially rename menu items. Let this test run for a few weeks at least to collect ample data about the user behaviour and their journey on your site.

Content could also be an issue: should the content leading to the category pages not be relevant and useful to the potential customer, they will not click-through to these pages. Google Analytics Behaviour Flow report shows the path users have travelled from one page to the next and will you show a percentage of where, in the user journey, visitors are dropping off.

It’s normal to see a large drop off rate from the homepage as not everyone who lands on the site is going to instantly buy from you. However, if you filter via entrance page and notice that users are consistently dropping off immediately from category pages then there is a problem. There could be a disconnect between what was said in the SERP (search engine results page) listing and with the content on page, or even that the content just isn’t compelling enough to ease them into purchasing from you.

Making a decision

It might be hard to find a statistically significant result when A/B testing. It’s a good idea to ensure you don’t make any changes unless the result is really convincing. For example, if your current conversion rate is around 5% but you know that your sector’s, or other collateral average, is 10% you should only make a change if it takes you within 10% of this final target. This would mean working to within a 90% level of significance. If conversions only goes up to 6-7% during the period you are testing a newly optimised page it may well be that this is a fluke, and conversion rates may well go down below the previous figure. You should be aiming for a minimum of 9.5% improvement in that situation.