3 Steps to Conversion Optimization Glory
Definition to implementation now. Weíll give you a three-step guide to easy conversion optimization testing:
Step 1. What do you want to test and why?
Define what it is you want to test and why. Testing for testing’s sake is bad business, so unless you have lots of spare time, you will want to find a current weakness or deficiency in your website or process and define the outcome you want to have (i.e., increased product conversion, increased sign up or contacts, etc.). Next you will want to benchmark your data. Take a snapshot of your analytics so that, in addition to your testing tool results, you will have your own metrics to compare against.
Step 2. Select testing method and design the test.
Once you have your test subject and goal you can select the testing method. Keep in mind the benefits and limitations of A/B and multivariate testing, especially traffic requirements and time requirements while you are reviewing your benchmark data from step one. Do you have good traffic levels (as in, several hundred or even thousands of visits per day)? How many variables do you want to test? How important is it for you to understand why the specific result was achieved?
An example is that you are an ecommerce company that sells backpacks. You think that your backpack sales are underperforming at a 2% conversion rate. You want to get some quick results on increasing clicks on your “add to cart” button. Your theory is that by changing the button from red to green, you may get more clicks and conversions. In this example, A/B testing would be a great way for you to test a single variation of your “add to cart” button: You’ll find out if changing the color from red to green, the only variable, will have an impact.
Step 3. Measure your results and start again.
Once you have a theory and a goal, and you have designed your test, you will want to watch/collect test data. Most testing platforms will have a built-in confidence level that needs to be achieved for a successful test. When this occurs and a winner of your test is declared, you can then move on to another test. Use what you have learned from the analysis of your step one. Were there other weaknesses in your funnel that could be tested? Are there any further complex multivariate tests you would like to take on? By making conversion optimization testing a standard part of your business, you are ensuring that your process wonít get stale.
Top Testing Ideas by Business Type
The best source for top-testing ideas will come from diving into your own analytics. Sort through your conversion funnels, landing page flows and cart pages. Look for drop off, site exit, bounce rates or other KPIs (key performance indicators) for your specific business. Below are a few ideas to get you started.
These Are Common B2B Tests to Consider
Contact or Quote Submission Buttons
Phone Number Placement
Call to Action Headlines (e.g., “get quote now,” “contact us for a free quote,” etc.)
These Are Common B2C Tests to Consider
“Add to Cart” Buttons (size, color, location)
Email Sign-up Form
Product Image Size
Shopping Cart Page
Call to Action Headlines (e.g., timed offers, urgency, short vs. long titles, etc.)
So then, what is the difference between multivariate testing and A/B testing? Multivariate testing shares the same baseline function as A/B testing in that you will be live testing element variations until a statistical significance or confidence level is achieved. The difference is in the deeper results generated from the test. While A/B testing measures a singular overall outcome (e.g., B version converted more frequently than A), you will not know if it was the different text, larger image or other element variation that caused the outcome. Multivariate testing will generate a test variation for every combination of possible elements you specify and present them to users/visitors. With this approach, you will be able to measure what variation combination has the most positive (and negative) impact on your goal metric.
As an example, you want to test your ad headline and product image to find out what combination will lead to an increase in sales. For each of these elements you’ll want to test two different headlines and two different images. The formula then looks like this: 2 elements (headline, image) x 3 headlines (1 control, 2 test) x 3 images (1 control, 2 test) = 18 pages in the test. You can see that this method would create many new test pages.