| If you can measure it, you can improve it. That is the | | | | has drawbacks. Number one is that you need |
| maxim of online testing. Those who test their | | | | thousands of impressions and hundreds of |
| advertising campaigns can see ROI double, or even | | | | conversions or click-through actions in order to |
| triple. Testing will increase conversion rates. There is | | | | achieve statistical confidence. The other setback is |
| no such thing as "good enough." You should always | | | | the ability to test only one element at a time. Testing |
| be testing and improving your campaigns. | | | | more than one element can take a lot of time and |
| Every element of an online marketing campaign is | | | | traffic. |
| measurable. Test your PPC ads, landing pages, and | | | | |
| email campaigns. Testing results in actionable data | | | | Multivariate Testing |
| that will show up in your bottom line. Split testing and | | | | Multivariate testing is the testing of more than one |
| multivariate testing are the two main forms online | | | | element or variable at a time. This form of testing |
| testing will take. | | | | allows you to test several elements at once and find |
| | | | the individual elements that bring the best results, as |
| Split Testing | | | | well as the combination of elements that will give you |
| Split Testing, also known as A/B testing, is the | | | | the best possible outcome. It allows you to find |
| simplest form of testing. Google AdWords, Yahoo | | | | which elements matter and which combinations are |
| Search Marketing, and most providers of email | | | | strongest in a relatively short amount of time. |
| marketing services have split testing capabilities built | | | | Multivariate testing can give you the best bang for |
| right into their campaign interfaces. | | | | your buck, and the testing algorithm automatically |
| Split Testing is randomly dividing visitors into two | | | | determines the best page design for all possible |
| groups and showing each a different version of your | | | | combinations of elements. |
| PPC ad, landing page, or email. The objective is to | | | | Testing at this level requires sophisticated statistical |
| see which version performs better in terms of | | | | analysis. Unless you are a coder with and advanced |
| conversion, CTR (click-through-rate), lead generation, | | | | math degree from MIT, you are better off using an |
| or some other target. | | | | off the shelf application or provider. Most providers |
| Email campaigns are easily tested. Send one version | | | | of multivariate testing are proud of their work and |
| of email to half the leads on your list, and another | | | | cater to enterprises with deep marketing pockets. |
| version to the other half. Use a subset of your list | | | | The exception is the Google Website Optimizer. It |
| for testing, and then apply the results to the rest of | | | | provides all the power of these other applications but |
| the list. You can use split testing in this way to find | | | | beats them on price. It's free. |
| the best layout, ad copy, and headlines for you email. | | | | |
| | | | Conclusion and Resources |
| Split Testing an AdWords Campaign | | | | Testing takes the guesswork out of marketing |
| Google AdWords makes split testing of ad copy | | | | campaigns and allows you to make continual |
| easy. Simply create two different ads to driving | | | | refinements that will increase your ROI. Implement a |
| traffic to the same page and see which one gets a | | | | thoughtful testing program and you'll be amazed at |
| higher CTR. It is important to turn off the optimize | | | | what works and how little things can make a huge |
| feature in the AdWords campaign settings so that | | | | difference. But most of all you'll be amazed at the |
| both ads get an equal number of impressions. | | | | difference it can make in your bottom line. |
| Another way of to AB test with AdWords is to | | | | See my article on increasing conversion rates with |
| have the same PPC ad go to two different landing | | | | Google Website Optimizer for more testing tips. |
| pages. This allows you to split test elements in the | | | | GrokDotCom has a wealth of information and |
| landing page for conversion or some other | | | | resources on the topic of online testing. Be sure to |
| performance indicator. Make sure the two pages are | | | | see their 7 free Google Website Optimizer resources. |
| significantly different if you use this method. Search | | | | Another good testing resource is Marketing |
| engines frown on duplicate data and may penalize | | | | Experiments. There you will find online testing case |
| you for it. | | | | studies and research. |
| Although split testing is simple and effective, it also | | | | |