We recently had the opportunity to work with the good folks at Barney’s Labels to execute a set of split tests on their website.
The purpose of the test was to experiment with different marketing messages, and then see which message did a better job of convincing visitors to submit a quote request. Barney’s is a print brokerage business and their website is designed to generate leads.
We’ll look at the details of how to run a split test in a moment, but first I want to talk about the changes that we made and the outcome that took place. Our experiment involved coming up with 3 alternative headlines to place on the homepage of the Barney’s site. We also experimented with alternative placements for the lead generation form that prospective customers use to make contact.
As you’ll see from the pictures below, we ultimately moved from a “brand” based headline to “benefit” based headline and we bought the lead generation form to the front page. The result was a 183% increase in the number of leads generated by the website. Prior to the test, approximately 2% of visitors would register their interest in buying printing services, and at the end of the test this number had moved to between 5 and 6%.
This was an incredible increase in leads for small business that relies heavily upon the Internet for new business. I think is worth noting, that Garry, the owner, is an astute businessman, who has spent many years streamlining his business and marketing processes. Indeed the website, created by another firm, was the recipient of a design award. What I find it fascinating that such a significant improvement was achieved with relatively minor website changes.
The magic is in the testing.
So what is split testing? In its simplest form, it is the process of devising multiple variations of your sales message, and in the case of the Internet, using software to randomly display these variations to different users as they visit your website. The software used to run a split test is reasonably intelligent, in that it will generally display the same test variation to a specific visitor if they visit the website on multiple occasions. The result is visitors are generally unaware they are part of a test.
The goal with testing is to find which variation reliably produces the best sales outcome. In our experience, especially when testing a website that has not undergone previous tests, an increase in sales or leads of 20% or more is common.
I personally regard a willingness to test as key indicator that the person I’m talking to is a professional Internet marketer. Testing requires a degree of humility because you acknowledge that you don’t know what sales message will generate the best result.
I can certainly testify, that in our own Noble Samurai business we’ve seen sales increases in the order of 200% through a series of split testing campaigns. Now I need to be clear, we have executed a number of tests that yielded no improvement, in fact this 200% increase in sales has required approximately 10 split test campaigns. However it’s been well worth the effort, with an estimated increase in revenue of conservatively $1.5 million pr year.
Testing can be do any time you ask a visitor to take specific action. For example, when you ask a visitor to fill in an online enquiry form, to buy something from you, or register their interest. Just about any measurable action can be tested.
So what are some of the benefits of testing?
Well apart from the obvious increase in sals it is worth noting the following benefits.
- Large Profit Jumps – most businesses can expect a large profit increase with even a modest increase in sales. This is because for most businesses, overhead costs don’t change a great deal as sales volumes increase.
- Advertising Become Affordable – as the performance of your website improves (we refer to this as improving your conversion rate), you’ll get more response from the advertising that drives traffic to your website. Many business owners report that advertising which previously was unaffordable becomes profitable as a result of the testing work they’ve done to increase the conversion rat on their website.
- Increased Affiliate Sales – for those businesses marketing themselves via affiliates, an increase in conversion will make your offering more attractive to affiliates who are paid a commission on sales they generate your website.
What should I test to increase the sales or conversion rate of my website?
The following is a list of common aras to focus on when devising a testing campaign:
- Big Changes – moving from a long form sales letter to a video sales message is an example of a large change. Typically you will want to test a range of largely different sales approaches, pick a winner, and refined from there.
- Focus on Benefits – most websites make the mistake of talking about the features of their product or service rather than the benefits. Simply altering the words in a key part of your website (e.g. headlines of your high traffic pages), to change the focus to benefits, will generally yield a significant jump in conversion.
- Use Some Known Human Influence Factors – authors such as Robert Cialdini have studied the field of human influence in depth. It turns out that there is a relatively small set of techniques that humans use to influence one another. From the perspective of writing a sales message some of the key influence factors include:
- Social proof – showing how others have benefited from the use of your service or product, e.g. testimonials, data feeds from social media.
- Scarcity – creating the situation where people will miss out if they don’t act quickly.
- Specific Claims – is has been shown that specific claims will typically outsell vague or general claims by a good margin.
- Risk Reduction – tactics such as offering an unconditional money-back guarantee has been shown to greatly increase purchase rates. In almost every case a money-back guarantee will yield a higher profit, even when the cost of product returns is considered.
- Clear Claims – finally, many websites are overly cluttered and the core sales message is drowned out by competing priorities. Simply by deciding what the primary action you wish your customers to take, and focusing the structure and sales message of your website around this, you will boost sales.
So how can implement a split test campaign?
In order to implement a split his campaign you will need to make use of testing software and have someone with basic development skills execute the test. We recommend, and primarily use, Google’s “Website Optimizer” testing software. I’m also quite fond of reedge.com although their software, unlike Google’s, requires a paid subscription.
In most cases the technical work to implement split test is simple. We’d be happy to help, and generally speaking, our testing campaigns cost less than $2000 to run. If your website is a core component of your business, it’s small, once off investment for a potentially large, ongoing gain. Alternatively if you already have a web developer, simply direct them to read the documentation for the software you choose to use.
Any other resources should I look at?
I’ve written a four-part blog post series over on the NobleSamurai.com blog all about conversion. As part of that series I give away my personal website conversion checklist that you can use. For those really interested I also suggest marketingexperiments.com as a fantastic resource.