Newport Interactive Marketing

Simple Tests to Improve Your Website

Which Test Won logoWhichTestWon publisher Anne Holland’s quiz-show-style presentation at February’s Newport Interactive Marketers focused on how to use simple tests to improve website sales, signups, and conversion rates.

Holland provided copious examples of how small – often nontechnical – tweaks can result in huge improvements. Here are the highlights of the presentation.

Why Synchronous Testing?

  • Create two versions simultaneously
  • Traffic engine splits it into two different versions to see which does better
  • Super cheap and easy to do
  • You can go to Google Experimental Engine or Visual Website Optimizer
  • Great for an agency: You’ll impress your clients

Why You Shouldn’t Go With Your Gut

WhichTestWon’s founder Anne Holland says: If you’re just using your gut you are probably wrong.  “I’m still surprised by the results,” she said. “I follow my gut, knowing what I know, and it’s the opposite. If my gut is wrong, yours can be wrong.”

Knowing the product or the client too well can actually be a drawback. You’re not the visitors, so you don’t look at the product the way they do. For this reason, you’ve got to test. Amazon tests, LinkedIn test, Dell tests, and you should, too!

Where to Start Testing

Do not start testing on pages with high abandon or bounce rate because you won’t see a much change.

Pages to Test & Track

  • Thank you & receipt pages
  • Follow up
  • Emails
  • Digital Marketing Campaigns
  • Landing Pages
  • Search Results
  • Category Pages
  • Product offer description pages
  • Abandons

How to Decide What to Test

  • Use the same size screen that your traffic will be viewing: a 15” monitor, 1-24x 786 resolution (Hint: 980 pixels across, 520 down)
  • Quickly glance at page to see what you notice that needs to be changed
  • Remove clutter
  • Vary headlines
  • Big images, change images
  • Buttons: sizes, colors, wording

Make One Change at a Time

  • If you test two changes on the same page, your results won’t be useful
  • Don’t change the headline and the image because you won’t know which worked better
  • Example: If you change the size of the button, don’t change the wording on the button

Or Test a Radical Redesign

  • Radical redesign means you make the dramatic changes.
  • Then you compare your old page to the redesigned pages and see which does better
  • Then tweak the winning page to see which elements made the difference

What You Want to Achieve by Testing

You don’t want just more clicks. If extra clicks don’t increase sales, no one cares.

Instead of simply more clicks you want

  • More sales
  • More qualified leads
  • Lower cost per lead
  • Higher revenues per cart

How Long Should You Test?

Typically you should run a test for 21 days

Testing Examples:

1) Dell Mega Image Test

  • Mega Image Won (Mega Image is an image that fills the majority of the homepage screen)
  • Increased leads generated by 36%
  • Lowered the page bounce rate by 27%

2) Dell Search Box Test

  • Test between a button that says search or a magnifying-glass icon
  • Search button won
  • Resulted in a 6.3% Lift in RPV (revenue) avg.
  • A winning test is usually results in a 3% lift

3) Local or Toll Free Number Test

  • Toll free number or local phone number
  • Which would generate more sales?
  • The local number increased sales (checkout page visits) by 18.3%

4) Email Test

  • Should it say: “Click here for more info” or “Find out more”?
  • “Click here” won
  • 97% lift in unique clicks
  • 16% increase in page views

5) Overlay Test

  • Overlays are like pop-ups on your website
  • They are much more responsive than saying: Please join our email list
  • Often come free with email service provider or free WordPress plugin
  • Test was how wide should the overlay be
  • Smaller overlay won
  • Results 8.8% More Opt-Ins

Don’t miss  Newport Interactive Marketers’ FREE monthly networking-learning gatherings.