When we have the opportunity to analyze a paid search account for the first time and are asked the question, "How can you improve my performance?" we look to landing page testing immediately.
Here's a quick illustration of the impact that landing page testing offers:
Increasing your conversion rate by 0.5 percent through landing page testing might not sound like much of a breakthrough, but the bottom line impact can be tremendous.
In the above example, the advertiser knows that a qualified lead is worth $35 to the business ($35 * Leads = Lead Value). A half percent increase in conversion rate leads to a $17,500 incremental revenue lift for a single month -- without an additional media dollar spent. Project that out over 6, 9 and 12 months and you can quickly see why conversion optimization and landing page testing are so vital.
There are numerous ways to execute an A/B split serve test. Server-side rotation, where you send all traffic to a single URL that then redirects traffic to test pages based on the design of your test, is going to offer you the most control and flexibility.
In this scenario, you will typically have a more even 50/50 split since you are controlling the rotation. You will also have more control of the dynamic variables that populate the test pages (dynamic phone numbers for example.) It does require a more sophisticated setup and added technical resources though.
If you don't have the resources at-hand, Google offers an alternate way to execute an A/B landing page test. The feature is called Website Optimizer, and it allows advertisers to quickly and easily set up landing page tests without the need for an intricate technical setup.
Before we walkthrough the steps of setting up a test within Google Website Optimizer, let's cover some things that you will want to think through as you plan your test.
Where to Make the Impact
If you have just taken over a new paid search account and are trying to figure out where to begin your quest for increased conversion rates, consider starting big. The landing pages that receive the largest proportions of traffic are commonly the pages that will yield the greatest return through testing. These pages offer the ability to impact your overall program significantly.
But what if your test page performs badly? Won't that bring down the performance of your entire program? It can if you aren't careful.
The good news: even when testing high trafficked landing pages, you can minimize the account's risk by sending only a proportion of traffic to the test, while sending the rest of the traffic to the standard control page. Later in this article we'll talk about how to do this within Google Website Optimizer.
As you prepare for your test, think about the keywords that will drive traffic to these pages. Look at the ad copy within the ad groups that these keywords are housed.
Are you setting the right expectations upfront with the consumer?
The ads that consumers see should always complement the landing page you are sending them to. Deliver on whatever promise you make in the ad copy.
Ensure that there is consistency between what you are saying within your ad and what the consumer is seeing when they get to the page. If "Free Brochure!" is in your ad copy, it should be prevalent on the landing pages.
Accept Failure, Move On
Mentally prepare yourself (and your boss or client) that the page you test may in fact perform worse than your control page. It happens, and it's OK.
Thomas Edison failed thousands of times before inventing the light bulb. Accept that failure may happen, minimize your risks, and be persistent in testing until you find the right set of attributes that lift conversion.
Don't Beat a Dead Horse
If you run a series of 14 tests based off of an initial control page, there's a good chance that you'll experience the largest conversion lifts within your first seven tests, and much smaller incremental gains from the last seven.
Every gain counts in conversion rate testing, but understand that diminishing returns often come into play here. Always continue seeking pages that offer the greatest upside to your overall program.
What to Test
Once you have your control page identified, we have seen the most success when creating a completely new test page layout. Change things up on the test page.
Consider differentiating these attributes on your test page:
- Headline and call to action
- Number of form fields
- Phone number placement (if applicable)
- Hero image
- One column vs. two or three column design
- Text and color of the form's "Submit" button
The number of attributes that can be tested are countless and deserve an article itself, but these are some good starting points.
A basic A/B landing page cadence could look something like this:
In the above example, Test and Test 2 could be quite different than the Control page from a layout and messaging perspective. The "iterative" testing (testing single attributes between the two pages) would then be conducted once we have identified the optimal structure and layout of the page via the first two steps above.
Google Website Optimizer: The Mechanics
Now that we've gone through some of the strategy and design of the test, let's get into the execution.
- Log in to your AdWords account, and select "Website Optimizer" from the "Reporting and Tools" tab. Click "Create a new experiment."
- When asked what type of experiment you would like to create, select "A/B Experiment."
- You will then be presented with an A/B Experiment Checklist explaining that you need a Control page and a Test page, and also a reminder to have your Conversion Page URL handy. Just check the confirmation box and hit the "Create" button.
- Now you'll be taken to the screen where you will input your original (or Control) URL, your variation (or Test) URL, and the conversion (or Thank You) page.
- Google will now generate the tracking scripts to be used. You will be given a unique snippet of code for your Control page, Test page and Confirmation page along with placement instructions for you or your web developer.
- Once these tracking scripts are placed, Google will validate that they are functioning properly.
- It's time to start your test! Here you will be presented with the URLs for the test and Conversion page, and the ability to "Preview this experiment now." Click this option as a double check that the proper pages are showing. Then, hit the "Start Experiment" button.
- We talked earlier about managing our risk when testing. This next step gives you the opportunity to do just that. Here you can set the percentage of traffic that will be sent to the A/B test. If you select 100 percent, all traffic will be sent to the A/B test. If you select 20 percent, than 20 percent will be delivered to the A/B test, and 80 percent will continue going to the original control page and not counted towards the test.
- Your test is officially live. You can go back to the main "Website Optimizer" page from the "Reporting and Tools" tab to edit your test settings if needed, and to view the results of your test.
- Within the report, you will be presented with the data needed to make a determination as to the winning page. In addition, Google will display a color-coded data scale to help in determining the winning and losing page in your test. Remember to be patient. It could take a few weeks of testing before you reach a statistically significant data sample.
Landing page testing is a combination of art, science, discipline, and perseverance. Over time, the payout can be big so be relentless in your pursuit of an improved conversion rate.
Save up to $400! Register now for SES New York 2011, the Leading Search & Social Marketing Event, taking place March 21-25. SES New York will be packed with 70+ sessions, multiple keynotes, 100+ exhibitors, networking events, and parties. Learn about PPC management, keyword research, search engine optimization (SEO), social media, local, mobile, link building, duplicate content, multiple site issues, video optimization, site optimization, usability, and more. Early bird rates expire March 4.