If you followed the recommendations in part one of this series, you’re now convinced that testing is important. And you’re committed to relying on data instead of listening to HIPPOs (the highest paid person’s opinion) when it comes to your online presence.

Congratulations on your new point of view. Making modifications based on test results can generate amazing success – in the case of one of our clients in the insurance industry, modifications resulted in a 70 percent lift in conversion rates.

So, where do you begin?

To get compelling results, there are a series of five steps to follow.

1. Start with a Landing Page

First off, you’ll need to find a page to test. It generally makes sense to start with a landing page – or a series of landing pages if you’ve got the bandwidth. This is an ideal jumping off point because by definition landing pages are more focused and have a single call to action – making it easy to isolate conversion variables and conduct tests.

Ok, makes sense, you say. So which page? Start with a page where key events happen that drive your business; a page that drives conversions.

2. Look for a Measurable Action

Next step, quantify exactly how this page drives business value. What do people do on this page that that aligns to your business goals? We’re looking for a measurable action. For instance, ‘fill out a form’ is a good example; ‘generate excitement’ is not. Some examples of landing pages that drive business value include:

  • An insurance company’s web page with an application form
  • A newspaper’s web page with a subscription form
  • An organization’s web page with an offer to sign up for membership

3. Review the Data

After that, look at your existing analytics data to see if the goal for this page is being met. Is there room for improvement? If it’s already performing well, you may want to look to another poorly performing landing page for a first test.

Or start with your most important page – the one from which you want the most conversions. Even if this landing page is performing well, getting more conversions is always a good thing. For instance, our insurance company customer decided that a 10 percent conversion rate on its application page, while good, still had room for improvement.

4. Look for Sticking Points

Now comes the hard part – you’ve got to analyze the page for friction and anxiety.

Friction is when people don’t know how to do what they want to do on the page. For example, users can’t find the call-to-action button, the form or the information they want to download.

Anxiety is when they know full well how to do it, but they don’t want to. They might not trust what will happen if they hand over their contact information or understand why you need a credit card for what they thought was a free transaction.

The analyst at our insurance company speculated that fewer site visitors might be filling out the plain-looking form because it asks for very personal information. That’s information people might not associate with their company, a trusted brand.

5. Make a Hypothesis

Once analyzed for friction and anxiety, you have an idea of what might be going on with this underperforming page. Simply turn that idea into a hypothesis: an if/then statement. For example:

If we make our brand more prominent on this application form, more people will fill it out.

Ready. Set. Go!

NOW we can begin to test. We have a landing page identified, an idea of what we might want to change, and an outcome we’re hoping for.

In our insurance company example, they came up with several changes to the style sheet and logos (large, small, absent) in order to make the page better match the rest of the site and more clearly associate it with their brand.

Come up with several alternate designs, or changes to the page that might influence your outcome. Don’t restrict yourself to minor refinements of designs you expect to be successful. Go for big changes.

Keep an Open Mind

Designers are often surprised by which option actually improves the outcome most. Give yourself room to be amazed. In our insurance example, every single design was notably less effective at driving conversions. (How’s that for a surprise?) Users preferred the bare bones form to any alternates.

Finally, let this be the beginning of a cycle of iterative improvement. There isn’t a part of your site that can’t be improved.

Our insurers came up with an alternate hypothesis based on the results of their test: Users want simple forms. “Slick” forms aren’t trusted.

After testing this they found that every alternate that simplified the form was more effective at driving conversions. They saw a 70 percent lift in their conversion rate once they applied the most effective combination of these simplified elements.

Now, if that isn’t reason enough for you to start testing, then I don’t know what is.

This post is Part 2 of a series on analytics to optimization – discussing how analytics can guide the most insightful MVT and A/B tests