Testing SEO Landing Pages and Improving Conversion

Testing has become a key online marketing discipline for increasing conversion rates and getting more out of search/pay per click traffic. There are several books on the topic, along with specialized software programs and companies that do nothing but help improve conversion rates through testing. If you think you need to start doing some tests yourself, you’re probably right; keep reading.

In this article I will introduce the concept of testing and how it applies to search engine optimization. Topics covered in this article an introduction to testing, SEO and testing, elements you can test and testing post landing pages.

The Biggest Companies Succeed Through Testing

Some of the biggest testers in the world are Amazon and Google. In fact, testing is one of their “secret weapons.”

Amazon used testing to succeed and continues using it to maintain its leadership in online retail space. The company started back in 1995 when consumers feared lightning will strike their heads if they pressed the wrong button. On top of that, everyone was extremely cautious about giving out credit card numbers and addresses over the Internet; e-commerce was still all but unheard of.

Yet through smart testing Amazon was able to survive and become a mighty player in the online space. Their strategy was (and is today) to test small portions of the website, without major overhauls, and find the versions that work the best — and then do it all over again. There is no “final test” for Amazon, as the company is on a 24/7/365 cycle of continuous testing and improvement. Consumers might not even notice it as the company tests different font colors, size and viewing options of images, position of images, calls to action, checkout buttons, etc…

A simple search for “Amazon test” on Google returns 6,590,000 results with the following headlines:

  • Amazon tests mail-order catalog listings

  • Amazon tests shopping-portal strategy

  • Amazon tests question/answer site

  • Amazon tests on-demand database

  • Amazon tests AdSense alternative

  • Amazon tests new mapping service

  • Amazon tests catalog sales

  • Amazon tests online grocery shopping

  • Amazon tests sales of sporting goods

  • Amazon tests new "leave seller feedback" page

{mospagebreak title=Even Google Tests}

Google is no newcomer to testing as well. It is known for extremely long beta periods, during which it rigorously collects and analyzes usage data, looking for information it can use to improve the product.

For instance, Google Wiki search sat for at least a year in Google Labs, before being released to the public. The actual public release was partial, as Google targeted a select user base before unveiling the product to everyone. Other products such as Gmail, Website Optimizer, Analytics, Ad Planner and more went through similar stages.

One Head Designer at Google had to quit his job over testing – here’s a quote:

"I had a recent debate over whether a border should be 3, 4 or 5 pixels wide, and was asked to prove my case."

In essence, the only way to find out anything on the Internet is to test it. The concept is as old as science. Have a theory? Go test it. If it doesn’t work, then come up with something else and test again.

A word of caution: do not test for the sake of testing. You have to have a direction and know what you’re testing instead of throwing things against the wall and leaving what sticks. There’s no time for that. You want to be agile. Knowing what to test is more important than testing for the sake of it.

Search Engine Optimization and Testing

Testing on natural search is a lot trickier than with pay per click. You cannot swap entire copy, nor can you offer A/B – multivariate pages through the same link as you can with PPC / banners (I’ll cover A/B and multivariate tests later on).

You can lose your ranking if you change too much, too fast. There’s no way to test new copy without affecting traffic and rankings as with PPC ads, since the Google bot would see the new content and would have to update the index, which in turn would rank you for slightly different phrases. Generally, if your links are strong you will still rank, but there’s a big chance results will jump up and down.

{mospagebreak title=Elements You Can Test}

It costs nothing to test and you can test EVERYTHING, be it a dynamic site or a static HTML page. Here’s a list of elements you can test on your website:

Headlines. I recommend styling H1 titles with CSS smaller than H2s and using them for search engines, while H2s and H3 are for visitors, with less of a keyword focus and more emphasis on persuasion/calls to action. Also: ask questions in the headline, evoke emotion, use various fonts, sizes and capitalization. Use features, benefits (or a combination), even quotation marks.

Images. Test positions, sizes, quality, angles, zoom. Try labeling images with descriptions.

Copy writing. Do not swap an entire page of copy. Add and remove paragraphs, add sentences and always observe search rankings. It is better to entice users to click to another page (which you can change all you want) than to lose rankings and traffic. Test copy length and message. Try direct marketing, regular copy writing or more technical articles (depending on your customer). Change bullet point order and the order of the paragraphs in the copy.

Shopping Cart. Make sure you get a flexible shopping cart and test out a few before choosing one to keep. Test the number of steps in the actual checkout, pictures of the products in the basket, the ability to edit the shopping cart at different checkout steps, registration (important: make it OPTIONAL), payment options, contact information, links to the products/different website sections, security certificates, privacy policy and terms.

Forms. Test various field sizes, the appearance of the “Send” button (don’t use "submit," try different wording). Test placement of elements and text/label sizes.

Calls to Action. Try various wording and formats such as links vs images (links have been proven to work better). Test shapes, colors and size.

Guarantees. Think return policy, money-back guarantee, etc. Test different policies.

Colors. Test different page colors, font colors, backgrounds and layout themes.

Links. Test various ways of wording your hyperlinks and watch for the ones that work best. Try logical and emotional links. Keep in mind that anchor text affects how search engines rank your pages, so be careful when testing this.

Reviews/Comments. Add reviews and comments to see how they affect your conversion rate. Make sure to disable registration so anyone can leave their opinions.

USP (Unique Selling Proposition). This a concise and unique statement that summarizes your business and tells customers why you are the preferred choice. The purpose of a USP is to answer the question “Why should I do business with you?” Try a few USPs and see which ones work best (USP is also known as Unique Value Proposition, or UVP).

Here are some examples of good USPs. FedEx: "When your package absolutely, positively has to get there overnight.” M&M’s: "The milk chocolate melts in your mouth, not in your hand." Domino’s Pizza: "You get fresh, hot pizza delivered to your door in 30 minutes or less — or it’s free." Test various positioning of the USP. Try putting it in the image near your logo, in the headline and within copy text.

Video. Add videos in various locations to test whether they help increase conversion rates/engagement.

Pages. Add new pages to the website such as guides and reviews to help buyers narrow down their choices.  

{mospagebreak title=Post Landing Pages}

What is a landing a page? In the SEO world, it is the first page users see when they get to your website. It is the page that actually ranks on search results apart from the index. I usually optimize landing pages for specific keywords.

Since content, titles, and H tags directly affect search engine rankings, making changes to landing pages can be tricky. You do not want to lose rankings and traffic.

To avoid losing rankings and traffic, you can direct users to your “post landing page(s).” Those are page(s) that have no search rankings, so you can change as much as you want without consequences to your traffic. Post landing pages can be:

  1. Accessible from the navigation menu.

  2. Pages that follow landing pages in a linear flow (lander -> posting landing page). 

Users rarely use websites linearly (though we wish they did), but linear flow is attainable after the lander if you entice users enough. The way post landers work is simple. Entice users to click on a call to action which leads to the post lander. The call to action may be a promise or something else, but you have to get them to click.

Keep in mind that the actual lander has to be persuasive itself and you will have to test different calls to action just to get visitors to the post landing page, so this method involves two pages:

  • Testing a call to action on the landing page which leads to the post landing page. We cannot test too much on the actual lander due to effects on search engine rankings.

  • Testing the post landing page – here you have 100% freedom to test.

Run tests on landing pages to get a feel before moving on to post landing pages.

In the next article I will get more into the actual specifics of testing various elements on the page.

[gp-comments width="770" linklove="off" ]