Article - An Insider’s Guide to Split-Testing By Alex Cleanthous, Web Profits

An Insider’s Guide to Split-Testing

Alex Cleanthous , Web Profits - Accessibility 0 Comments | Add Yours

About The Author:

Alex Cleanthous is the Head Online Strategist of Web Profits, an Australian-based online marketing firm servicing Australasia, the US and Europe. He keeps Web Profits on the cutting edge in all aspects of online marketing. Download a free copy of our special report Website Conversion Strategies at www.webprofits.com.au

What Is Split-Testing?
Well before the Internet was created, direct marketing legends such as John Caples and Claude Hopkins would setup a split-run-test in a printed publication where they would test two different advertisements (for the same product) and measure the success of an each by using different coupon codes and measuring which version of the advertisement produced the greatest results – half of the publication would have one version of the ad, the other half would have the other version of the ad.

Fast forward 100 years and the same process is being used on the Internet by savvy website owners, referred to as split-testing or A/B testing.

Split-testing is the process of creating two variations of the same web page or advertisement, then splitting the traffic between the two variations evenly and measuring which version produces the best results. The first person would see version one, the second person version two, the third person version one and so on. The version with the best results is declared the ‘champion’ and all other split-tests are created with the goal of beating the champion.

With Google Analytics and Google Website Optimizer split-testing can now be implemented relatively easily by any website owner. The real challenge lies not with knowing how to split-test, but knowing what to split-test.

Benefits of Split-Testing
The most valuable benefit of split-testing is that you can achieve a greater return on investment from the same advertising spend. For example, let’s say you spend $5000 on a pay-per-click campaign and receive 5000 visitors to your website. Your website converts 1% of those visitors into sales, generating 50 sales for your business. By increasing your conversion rate from 1% to 2% through split-testing you would generate 100 sales for your business instead of 50, from the same advertising spend.

Split-testing also allows you to:

•	Compete in very competitive markets – where a high converting website is needed just to break even
•	Dominate less competitive markets – where a high converting website will capture the majority of the market share
•	Make PPC advertising profitable – where every dollar saved is added to your bottom line
•	Remove any guesswork or false assumptions – made during the design or setup process
•	Incrementally improve your returns over time – by continually testing and refining every element of your marketing
•	Improve the performance of your affiliate program – the better your website is at converting visitors the more traffic your affiliates will send you
•	Settle debates between stakeholders – if two decision makers in your business have differing views, don’t argue about it... just split-test it!

How To Split-Test Your Website
Split-testing requires you to have a success page that your visitors are taken to after they complete an action on your website, such as completing a form or purchasing something from your online store.

You can then use Google’s free Website Optimizer software to manage the entire process for you. All you need to do is open up an account with Google, register for Google Website Optimizer, setup the test and then paste some code into your website. Google then rotates your visitors between the two versions, tracks the results and then lets you know when you have a winner.

You should be split-testing all of your highest visited web pages all the time. A good idea is to set one day per month (eg the first Monday of every month) where you review all of your split-testing campaigns, identify the champions and then setup new split-tests to beat the champions.

It’s also a good idea to create a spreadsheet where you can record what you split-tested, the dates the tests started and finished, and the results. This ensures that you don’t run split-tests over and over again (because you forgot) and also allows you to identify patterns over time.

What Can You Split-Test?
One of the best things about the Internet is that you can measure (and therefore test) everything, which means you have no excuse for running a campaign that isn’t producing the results you want. The following components of your online marketing should be split-tested on a regular basis:

•	Website Design – your website is the most important area you can split-test as it affects the performance of all other online advertising. The better your website is at converting visitors into sales or leads, the greater return you will achieve from your marketing budget. You should be split-testing your headlines, offers, sales copy, images, testimonials, pricing and anything else contained on the pages in your conversion and sales funnel.
•	Pay-Per-Click (PPC) Ads – what your PPC ads say in them can play a very large role in the mindset your visitors have when they land on your website which, in turn, can affect the performance of your campaigns. You should be testing headlines, angles, offers, negative qualifiers and landing pages.
•	Banner Ads – banner advertising can be a very profitable source of leads and sales if done correctly, but it can be very difficult to find the right combination of factors for it to produce results. If you are running banner ads you should be testing headlines, offers, colour scheme and landing pages.
•	Email Marketing – according to research conducted by the Direct Marketing Association in 2009, email marketing generated a return on investment of $43.62 for every dollar spent on it. Achieving results like this will require split-testing of subject lines, offers, copywriting and landing pages.
•	Offline Advertising – like the direct marketing masters of the early 1900s you can run a split-run-test in your offline advertising and use coupon codes, tracking phone numbers or tracking URLs to measure the performance of your advertising.

To Split-Test or Not To Split-Test?
While website split-testing is one of the best methods of improving the return from all of your marketing, it may not be the first point of call. Before you can start split-testing you need a website that already produces results and enough people visiting your website so you can start and finish a split-test within 30 days. Google recommends that you receive at least 100 conversions (sales or leads) before declaring a split-test complete, so divide the number of conversions you are currently receiving each month into 100 to see how many months it will take you to complete a split-test.

Generally, you should be receiving at least 2000 visitors per month to your website before considering split-testing but you can do it with as much or as little traffic as you want – the higher your traffic the faster you will achieve results.

If your website is receiving traffic but not many conversions then consider a complete website redesign. While split-testing can definitely improve the performance of a website over time, a poor performing website should be overhauled by website conversion experts first, and then be split-tested.

You should split-test the rest of your advertising from the moment it is setup.

Getting Ideas For Split-Testing
As mentioned at the beginning of this article, the real skill in achieving results with split-testing is not knowing how to split-test, but knowing what to split-test. With that in mind, here are some recommendations to help you get ideas you can then split-test on your own business:

•	Check Competitors Sites – check your competitors’ websites regularly and see what captures your attention. Did they recently add a new logo? What offer are they using? What are they offering that you’re not? Competitors can be a great source of split-testing ideas.
•	Create A Swipe File – a swipe file is a file of good advertisements that have caught your eye. I personally use Evernote to capture good websites, landing pages, banners and ads that have captured my attention. You can then review your swipe file whenever you need inspiration for new split-test ideas.
•	Visit ABTests.com – this is a user-generated website with results from split-tests conducted by website owners, designers and conversion experts from around the world.
•	Educate Yourself On Direct Response – split-testing is all about direct-response (ie getting your visitors to respond to your offer now). Educating yourself in this area will directly benefit your split-testing efforts. A great place to start is reading the books Testing Advertising Methods by John Caples, My Life In Advertising by Claude Hopkins and Ogilvy On Advertising by David Ogilvy – even though some of them were written more than 100 years ago, the lessons you’ll learn are just as relevant today
•	Usability Testing – use online services such as UserTesting.com and FiveSecondTest.com for usability testing on your own site. Find out where visitors have the biggest issues and then split-test your website to resolve these issues.
•	Hire A Company To Do It For You – there are many companies that specialize in conversion rate optimisation (like us) and can help you optimize every part of your online marketing campaign.

The Secret To Successful Split-Testing
Never, ever make an assumption that what another company is doing will work for you, no matter how successful they are. Always test every idea you have and wait for the test to finish - you would be surprised how many times a split-test changed directions towards the end of the test. Finally, always be testing... we test every component of our online marketing all the time and our online marketing performance continues to improve. We’ve been doing it for years and we’ll be doing it for many more.