CHALLENGE
"We do advertising on TV, radio and online," says JustListed.com Senior Marketing Director Matt Heinz. "The TV ads have been running for over six years with a little cadence to our URL at the end. People remember the name pretty well. The URL gets in your head, it's been pretty powerful to build recognition."

Which is fabulous for branding and driving traffic to the site, but in the end the site's entire business model rests on one single factor: getting consumers to register for emails on new housing listings in their chosen area.

And as Heinz is well aware, these days it's tougher than ever to get consumers to register online. "Consumer behavior online is changing. There's more ambivalence toward forms. If you optimized your form a couple of years ago, it's likely your complete form may need to be changed."

Heinz had been running a/b tests to keep up with consumer tastes. But a/b tests have their limit when you consider the hundreds or even thousands of tiny tweaks you can test on a form to try to get conversions up.

CAMPAIGN
During the height of house shopping season this spring, Heinz decided to run a series of multivariable tests, testing hundreds of variations for 15 main form variables over three waves. (See link to samples below.) The three most important tests included:

Test #1. Order of questions on form

Heinz had already learned from past tests that he needed to clump his questions into just three sections to get the best results. These sections were:

Section A. Additional areas in which you would like to receive listings

The entry page to the form had already asked just one question, what zip code were you looking for a house in? This next section gave a long three-column list of up to 50 additional nearby towns and neighborhoods the consumer might consider.

Sounds too long, but past tests had determined the more options Heinz gave consumers at this point, the more value they assigned to filling out the form. They felt truly engaged and as though they'd get a valuable payback in the form of personalized information in exchange for taking their choice from this list.

Section B. Tell us about your current situation

This asked for email address along with data on when the consumer intended to buy their next house and if they'd pre-qualified for a loan.

Section C. Tell us a little more about what you're looking for

The consumer could take choices from three drop-down menus (square footage, property type, type of construction) and enter freeform comments about their housing desires into an empty field.

Test #2. Introductory copy

Heinz' control (the previously winning form from a/b tests) had only one sentence of introductory copy that ran two lines long at the top under the site logo:

"Please confirm the neighborhoods or cities in which you'd like to receive listings, as well as a few more details about your dream home."

Now he tested several variations on even shorter copy that ran only one line long, including: "Great News! We can also offer you FREE listings in nearby areas."

Test #3. Submit button copy and placement

Heinz's control used a button that read in all caps "SUBMIT." Now he tested several different wordings and tested placement -- would it make a difference if the button were in the center of the end of the page or should it be at the far right bottom corner?



RESULTS
The winning registration form looked very much like the original control; in fact, you have to look carefully for significant changes. However results were anything but insignificant -- conversions rose by 34.3%.

What made the difference?

Lesson #1. Changing the order of the questions to put the longest in the middle and start with what Heinz calls the "dream" question, "A little more about what you're looking for." He says, "Folks build a story in their minds, it's very aspirational. You get people on a roll, you want to keep them on a roll. Once you've described it, you're much more motivated to wonder if you could actually find it."

Worth noting -- asking for the consumer's contact information at the start of the form was *not* a winning tactic. Yet, it's the way most registration forms start out.

Perhaps form-designers should put the consumer's interest at the start of the form and the marketer's objective (an email address) near the end. That way you're getting power from the what's-in-it-for-me factor.

This is probably why registration forms from subscription Web sites such as eDiets, Match.com and Classmates *all* start with questions to help the consumer and only near the tail end of the registration ask for actual contact data.

Lesson #2. Shorter intro copy works better (or at least it's well worth testing). The one-line intro outpulled the two-line intro even though they were both just one sentence long.

Lesson #3. Test your submit button. In this case, just as with every other button test we've reported on, the winning copy was *not* the word "submit." Instead this time it was, "Get your FREE Listings."

This matched the wording in the winning intro copy (another factor we've noticed in winning buttons across several sites).

Naturally, Heinz plans to continue testing new variations on this page, so if you go to the site in future you may not see the current winning design. "There are a thousand things we could try on the site," he says. And he's looking forward to testing every single one of them.

Useful links related to this article

Before-and-after screenshots from JustListed.com's test: http://www.marketingsherpa.com/justlisted/study.html

Optimost -- the multivariate testing firm that JustListed.com used for the tests http://www.optimost.com

JustListed.com http://www.justlisted.com