Speaking in tongues: A very simple A/B switch for testing the pull-power of landing-page variations

by Greg Swann

Suburban Phoenix Real Estate Broker

Author website

We talked about landing pages at Unchained: When someone who is interested in relocating to Phoenix lands on our brokerage weblog, I want for that party to land on my relocation page, rather than just at the top of the blog. Why? Because if I provide the exact information my visitor is looking for, I have a much better chance of converting that person into a client.

This is important: Social media marketing is direct marketing — target marketing, not viral marketing. WordPress sells itself by viral marketing. You sell your business on a WordPress weblog by direct marketing, by focusing your attentions on particular, identifiable prospects. Of all the people writing on BloodhoundBlog, the on who has the most to teach us about this is Richard Riccelli, Delphic and sphinx-like but overflowing with brilliant direct marketing ideas.

So what’s better than a landing page? Richard can beat me up, if he wants, for getting this all wrong, but the direct marketer’s answer is simple: Better than a landing page are two landing pages — pitted against each other.

Advertising is a prayer to the heavens, but direct marketing is testable. Is long copy more effective that than a shorter appeal? Test it. Will a question or a promise work better as the headline? Test it. Do brief forms produce more leads? Do more rigorous forms produce stronger leads? These are testable propositions.

But: There is a caveat: You have to be getting enough traffic to make testing worthwhile. If your long copy beats your short copy two-to-one, it means nothing if you only have three instances to judge from. A Google Adwords campaign is eminently testable, as is a Zillow EZ Ads promotion. On your hyperlocal real estate weblog, you may have to let your tests run for a while before you draw any conclusions.

And what should you do when you do prove that one way works better than another? Test two variations of the winning strategy against each other. And test everything else while you’re at it.

Okayfine. There is no limit to what you can learn about direct marketing, and Brian and I going to do a ton on it in Orlando. Here are one, two, three Amazon.com book lists on on-line direct marketing. Your time is finite, I know, but this, like everything, is a leverageable problem: Focus your attention on the pages and forms that can make you the most money right away, then sweat the smaller stuff later.

That’s the what. Here’s the how. Yesterday in email, I suggested to Chris Johnson that he should be testing some of his landing pages at Loan Officer Survival Training. At his personal weblog, Chris said:

What I must do is make sure that EVERYTHING leads to one or two pages (set up for A/B testing as kindly  suggested by Greg, more PHP to learn), and that those one or two pages are the best representation of me–and what I want–that they can be, tested, vetted, refined, perfected.

Chris has probably already written the code he needs. This is a very simple problem in PHP. Here is a very simple routine to redirect inbound traffic to one of two testing pages:

<?PHP
if (rand (0,1))
	include ("TestVersionA.php");
else
	include ("TestVersionB.php");
?>

You can see this in action by clicking here. Just keep refreshing to watch it work; it’s a simple coin flip. The code shown above would have the URL of the landing page, and then the two (or more) testing versions would be uniquely named. The end user will see the same URL in any case. The pages you are testing will need some kind of key for you to compare their respective responses — a different email subject line, for instance, or simply a “coupon code” in the form response to distinguish one version from the other.

There are three kinds of web-based marketers who can profit from this kind of A/B direct marketing testing:

• If your current results are lousy, you can figure out what you are doing wrong
• If your current results are mediocre, you can figure out how to make them better
• If your current results are fantastic, you can figure out how to kick your own ass

Advertising at its best is religion. Marketing at its best is science. This kind of A/B testing of your persuasive tools is the threshold of marketing as a science.

 
Amending this: The code I’m showing above is workable and easy to deploy, but I don’t love it. This is almost as simple but more supple and more subtle:

<?PHP
if (rand (0,1))
    {
    ?>
    <!-- Version A: Insert any valid HTML -->
    <?PHP
    }
else
    {
    ?>
    <!-- Version B: Insert any valid HTML -->
    <?PHP
    }
?>

You can use that in any spot within a page where you want to test alternate versions of text or a form. For the landing page job we’re talking about, both versions of the page can be within the same one PHP file. But if you wanted to test alternate versions of a response form, you could do that test within an otherwise static page. This now becomes so easy that I will caution you against testing too many things at once. I could show you how to build a response key in PHP so you could see which one of your 32 possible variations scored — but I think that’s way too much testing to yield reliable results.

 
Further notice: Richard Riccelli sent me a note telling me I’m all wet, which is no doubt true, but here is a further refinement on this idea. My belief is that a strong net-based lead comes from someone who has taken some care to find out about us. So one thing we could do is lay down a cookie for each mission-critical page on our site (including testing variations), increment that cookie every time that visitor comes to that page a subsequent time, then report out that entire cookie history when the visitor submits a form. Over the long run, we would be able to correlate completed transactions to page views to find out who really is more likely to act — the ants or the grasshoppers.

Technorati Tags: blogging, real estate, real estate marketing, real estate training, technology