How To Pick A SEM Agency: Part 3 A/B and MVT Experimentation
For the past few weeks we’ve been looking at a blog series entitled, “How To Pick a SEM Agency.” The series is designed to help determine the components that go into The third of five part series will focus on A/B and MVT Experimentation, and how it plays an integral part of Search Engine Marketing (SEM) and should be a major consideration when picking a SEM agency or consultant. The rest of the series will focus on:
- Search Engine Marketing
- Web Analytics
- A/B and MVT Experimentation
- Landing Page Optimization
- Online Marketing
A/B and MVT Testing
A/B and MVT Testing in my opinion is one of the most neglected components of online marketing today. Very few individuals, agencies, and companies engage in the practice yet it is relatively simple and cost effective to setup and use. For the sake of this discussion we’ll be discussing testing as it relates to the landing page in which you’re driving visitors to. You can also engage in A/B testing to a certain degree in your ad copy, but we’ll cover that topic on another day. A/B testing is the testing of two variations of a landing page. You can change out any number of variables and variations on the page, but the software directs the visitor to either A or B. As a best practice you will probably want to make a small change of only one variable and one variation to understand what caused the difference in your goal. When do I recommend A/B testing? Simply put if you’re new to testing run an A/B test to get your feet wet. The second use of A/B testing is if you have a small amount of traffic going to this page. Mulitvariate Testing as I’ll discuss in just a minute typically requires large amounts of traffic to see results in a prompt manner. This description is greatly simplifying the concept of testing but in the future we’ll delve deeper into more advanced concepts.
Multivariate Testing is the testing of multiple variations and variables on a single landing page. In an MVT experiment you can change out items like the hero image, call to action, form, colors, etc. and see what the optimum combination is when it comes to achieving your objective. The more variables and variations that you have the greater the amount of traffic that you need. Simply put if you have ten visitors a day coming to your landing page it could take years to see statistically significant results.
Now how is this applicable to SEM? As with web analytics A/B and MVT experiments allow your SEM consultant or agency to further enhance your conversion rate by tweaking the landing pages in which visitors land to find the optimum combination. You may find through more complex MVT testing that visitors from Yahoo or Google perform differently on different pages. Similar results have been found down to the keyword level. Often times MVT applications can cost $50 – 150k a year or more but for businesses that may not have the budget necessary to make such a large expenditure Google Website Optimizer is a free tool and offers basic levels of A/B and MVT experimentation. If in the future you require more advanced testing options you can look into Optimost, Omniture Test & Target and Site Spect among others.
I’ve personally seen tremendous results and lifts in conversion rates from 100 – 600% increase depending upon the site and the client. Testing is something that your SEM vendor should be aware of and should offer as a professional service. Ensure that as you question any potential consultant or agency that you see if they have experience in A/B and MVT experimentation and what results that they have seen in the past. The concept of A/B and MVT experimentation deserves far more attention, but the concept with regard to this series is to understand the basics so that when you do quiz a potential SEM vendor you know enough to make an informed decision. For more on the concept see Jonathan Mendez’s blog as he is one of the foremost experts on the subject. If you have any questions about this post or any other blog posts please feel free to contact me.
A/B Testing, Multivariate Testing, Search Engine Marketing, Testing
Recent Comments