Can Crash Test Dummies Improve Your PPC Marketing?

General Motors understood the importance of testing.

Back in the 1970's, as the number of serious injury accidents increased amid smaller more fuel-efficient vehicles, GM was forced to find a cost-efficient way to develop safer vehicles.

Instead of waiting for the dreaded fatality stats, General Motors developed the first "crash test dummy" to proactively test various safety configurations.

Logically, it made sense to minimize financial risk by testing production changes before committing to re-tooling the entire manufacturing.

Logically, it makes sense to test your landing pages to minimize your financial risk by testing various messages, images and design before committing all of your PPC budget to one landing page.

Whether your web site is designed for Ecommerce or lead generation, why not be sure you're designing the most productive pages by using your own crash test dummies?

Landing Page Testing To Become Mainstream

With Google's upcoming public release of their free landing page testing technology for AdWords (called Website Optimizer), along with 3rd party vendor solutions from companies such as Vertster, Inc., there's no reason why landing page testing should not be a key part of your online strategy.

The upside potential to landing page testing can be dramatic improvements on your return on ad spend. The downside is the additional account management time required to design, measure and evaluate your tests. According to a recent study by the e-tailing group, 33 percent of you already spend at least 21 hours per week managing your PPC campaigns. With landing page testing, expect that number to rise.

Understanding Testing Methodologies

Now before we start creating our own PPC crash test dummies, let's get a good basic understanding of the available test methodologies.

1. A/B - testing two versions of a page against each other. The advantages of A/B testing are the relative quickness and ease of administering the test and the fact that you're measuring the actual behaviors of your visitors; A/B testing is very good at evaluating even minute differences in performance.

However, you can only test one element at a time, and can't see how the interactions of multiple elements work together to improve conversion or curtail response. Also be aware that repeat visitors may not respond as well to your modified test page since they may have grown accustomed to the look and feel of your original page.

In A/B testing, there are multiple scales of testing besides the traditional 50/50 split test. If you want to minimize the risk of losing business in a test, you can do a variable rate test such as a 90/10 distribution, where 90% of the traffic is sent to your control page and only 10% to the test page. In exchange you may have to exercise a little more patience waiting for statistically significant results.

2. Multivariate - testing multiple elements on one page. Multivariate tests are more flexible in that they allow you to simultaneously test more than one combination of modified elements (e.g., headline, copy, image, etc.). Multivariate tests usually can determine the impact each variable has on conversion alone and in combination with the other variables. These types of tests are more expensive, and depending on the testing method used, may predict the winner based on complex mathematical algorithms.

The 5 Rules of Successful Landing Page Tests

The ultimate success of your test is heavily influenced well before the first visitor lands on your new page. Here's a list of items that increase the chances of a successful test:

1. Your test goal should be a well-thought out process and in alignment with your overall marketing strategy. Be sure not to get distracted by the power of testing. Creative testing takes resources, and your hoped for long term results should easily generate a pay back many times your cost.

2. Ensure that your measurement strategy is done in advance, including implementing the right measurement tools and processes, as well as defining your success objectives and thresholds. If you're testing an Ecommerce page, ensure that your reporting will include product sales information. A test page could convert a higher percentage of visitors to buyers, however generate fewer sales dollars than the control page.

3. Have a clear understanding of potential test risks such as loss of revenue and down pages due to technical issues.

4. Landing page tests may cross over functional department boundaries within your organization, so make sure you have management buy-in, and you have efficient communication among all departments. There's nothing more frustrating (or embarrassing) to your sales people than to take a call from a prospect who has seen an offer unbeknownst to your sales people.

5. Define the various role responsibilities:

  • "Owner" of the test or Administrator
  • Creative design
  • Technical implementation
  • Tracking & Reporting
  • Final Decision Making

What Should You Test?

There are three primary influencers that impact the effectiveness of your landing pages: Your offer, creative, and messaging channel. Generally, it's a good idea to not let any pre-conceived notions or assumptions about your landing page define your testing strategy.

Primary Influencers in Landing Page Tests

Offer - The offer includes pricing, shipping and payment terms, product benefits, and copy, including your call to action. Here are some items you may want to test:

1. Discounted pricing - test dollars off versus percentage off pricing

2. Free shipping

3. Language in the call-to-action:

  • "Buy now", "Order Now", "Download Now"
  • Placement of guarantees

4. Language in the sense-of-urgency claims:

  • "Order by December 31st", "Quantities Limited"
  • Size and placement of text

Creative - The creative involves the page display including images, headlines, formatting and use of bullets. Common creative items tested are:

1. Placement of call-to-action

2. Use of visual cues, such as arrows to guide the reader

3. Use of text links versus buttons

4. Use of different colors

5. Testing different images, for example an image of product versus an image of a customer using your product or service

6.Placement of trust enhancer's on the page

  • Testimonials
  • Privacy policy
  • Trust marks

Messaging channel - The messaging channel consists of the source and the message contained at that source that generated the visit. Common sources include the various search engines, email campaigns and printed collateral. It's important to recognize that the results you get from one channel should not be interpreted as representative of the performance you'll get from all channels.

For instance, due to the differences in demographic makeup between typical Google users and MSN users, you should not infer that test results involving your visitors from Google will necessarily apply to MSN users.

The message displayed at that source is the other part of the messaging channel. For the pay-per-click engines, the message is your ad title and creative. Any differences in your ad titles or creative will also impact your test results, so you'll want to figure these tests into your overall test plan.

Even differences in visitor behavior should be anticipated depending upon whether your traffic originated from paid search or organic search ... even from the same search engine.

The take-away here is to be sure to structure your tests to be able to accurately measure your results based on the specific message channel driving your customer's visit.

Make It Easy To Get Started

There are a variety of metrics within your web analytics solution that'll help you identify which pages are the best candidates for testing. One of the best indicators that your landing page has room for improvement, is called the bounce rate.

A page's bounce rate is defined as the percentage of visits to that entry page where the visitor did not proceed any further into the web site. Since, in almost all cases, a landing page is designed to prompt a further action from the visitor (fill out a submit form, place a product in the shopping cart, navigate to another page), we can equate a bounce as a failure of our landing page.

Below is a bounce rate report from our Web Analytics Solution, Conversion Analyst. Notice that by comparing the bounce rates of the various product landing pages, we see that the Bathrobes and Slippers page has the highest bounce rate. About 1 in 2 visitors that enter the site on that page, leave without going any further. In addition, since this is our 8th most popular entry page, the outcome of the test should be significant enough to justify the costs of a test.

Finally, since the Bathrobes and Slippers page is one of a multitude of product pages, our test conclusions would be a good starting point for testing other high bounce rate product pages.

The Bottom Line

If you do any pay-per-click marketing, landing page testing needs to become an integral part of your strategy. As bid prices continue their slow ascent and your competitors test and optimize their landing pages, finding out what drives your online conversion has never been more important.

Many thanks to Scott Miller, CEO of Vertster for his invaluable feedback on this newsletter.



Bookmark and Share

Copyright © 1999 - 2016 Engine Ready, Inc. All Rights Reserved.