CHALLENGE


"Most trials come from the first link, so it's very important to get them then," says Bernardo de Albergaria, VP Marketing Citrix Online, on marketing GoToMyPC, the company's lead product to the SOHO/small business marketplace. "We have to put as much effort as we can into converting them in the beginning."

GoToMyPC's online ads originally routed anyone who clicked straight to the product's homepage.

The trial was free, but required users to enter a credit card number, and the visitor to trial ratio hovered at 0.3%.

De Albergaria knew conversions should be higher. So in 2001, he rerouted his ad clicks from the home page directly to the product's registration page to see if that would help.

It did. Conversions more than doubled, rising to 0.7%. "We were like, 'Oh, we got something here.' We lose fewer people by getting them immediately to where they want to go," he says. "But the registration page was built as a segue from the home page. It wasn't optimized to let people know what we were."

It was time to get serious about creating a landing page that would work.

CAMPAIGN


Because the company's media buying changes all the time, "we couldn't always customize the page, so we had to find something that worked for everybody," de Albergaria says.

Using A/B rotation capabilities developed in-house, his team started testing basic elements of the landing page (link below to 14 creative samples from various tests). Each test ran long enough to get statistically significant results. "At the beginning when we had less traffic, it took a couple of weeks. Later it took less time, three or four days." Testing time is also reduced in cases where there's a clear winner. "If we see something that is 20% better let's say, we assume that most likely it's going to be better, even if the margin of error is high," he says.

-> Step 1. First series of testing

a. Headline and call to action Because the winning headline from ads was "Access and Control Your Computer from Anywhere," the team carried that over to the landing page. They also tested "Access Your PC from Anywhere," and "Access Your PC from Anywhere. TRY IT FREE"

They added what de Albergaria calls the ugly burst, an orange starburst that says "Free Trial." And they strengthened the call to action, from "Get Your Risk-Free Trial" to "Register Now! Setup takes only two minutes."

b. Copy They tested short descriptions in bullet point format, longer descriptions, and benefit/usage examples rather than features.

c. Pictures They tested: --illustrations vs. photographs --color vs. black and white --people vs. no people --men vs. women

d. No links "We just want them to register [so we tested] giving them only one way to go," says de Albergaria.

e. Press quotes "We tried to build the page by including elements for the different personalities that might come to the Web site: for the one who knows what she wants and wants to get right to the trial, the form is right there; for the one who needs third-party validation, the quotes are right there; for the one who needs benefits, the copy is on the left," de Albergaria explains.

-> Step 2. Second series of testing

The trial flow for GoToMyPC is composed of several required actions. Users must enter their personal data, a password, and their billing info, then download the product. De Albergaria's team began testing the elements of that trial flow.

a. They brought the credit card registration right up front and combined it with the registration form.

b. Instructions on the form "The instructions on the fields were not clear. For example, ours is high security so the password has to have eight characters, including letters and numbers," says de Albergaria.

He tested putting the instructions in bold, with a green arrow next to the first field so users knew where to start.

He also tested the wording in the fields:
--"create account" vs. "start here"
--"password" vs. "new password" vs. "create a new password"
--"re-enter password" vs. "re-enter new password" vs. "re-enter your new password"

c. Adding links He also added a second page with more information, and links to pop-ups.

"This was getting us closer to the concept of a minisite without navigational items to the [homepage]," he says. "It was a tunnel page, you could only go back and forth."

Links included: --From the press quotes, de Albergaria tested links to the entire article, "even if it wasn't entirely positive," he says. --How it Works --Features and Benefits

d. Button style "We tested everything from regular gray system buttons to different styles and colors," de Albergaria says.

e. Pricing He tested putting pricing information right on the front page

f. Registration flow "Because you lose people with every step, we tried to keep it as short as possible," he says. They tested combining all the registration requirements into a single form (first name, last name, email, password, credit card), two forms (first, last, email, and password, then credit card), and three forms (first, last, and email, then password, then credit card).

The team also tested removing the necessity of replying to an activation email (where users would have to confirm their address via email before they could complete registration).



RESULTS


Conversions increased almost 400% in the first series of testing, and increased again in the second series. Specific data:

- "Access Your PC from Anywhere. TRY IT FREE" was the winning headline.

- Short bullets worked better than longer format copy, fewer links increased conversions, and press quotes increased trust.

- Pictures made a big difference. "At the beginning, female worked better than male, black and white worked better than color," de Albergaria says. "But then we started experiencing diminishing returns when we added the next round of improvements. No matter what kinds of pictures we used we didn't see major improvements."

- Surprisingly, when the credit card registration was brought up front, conversions dropped by 50%. "It scared people, without having already sold them on the benefits," he says. "It's like going into a store and instead of the clerk saying, 'Hi, how are you?' they say, 'Hi, make sure you have a credit card.'"

- Bolding instructions on the registration fields increased conversions by 15%.

- "Create account" increased by 10% over "start here."

- "Create a new password" showed a 20% increase over just "password."

- The size, style, and color of buttons had significance. "We tested everything from regular gray systems buttons to different styles and colors," he says. "We found that size was more important than style and color. Right now, green is working well." He also says that it's important their buttons look "clickable" and not too fancy.

- Adding pricing upfront depressed results. "We took that out. They would find out about pricing later in the process."

- Counterintuitively, the three-step registration process (first name, last name, and email, then password, then credit card) worked better than a two-step process. "We do not know why, and we don't spend time and resources trying to find out why," he says. "If it works it works, and if it doesn't, it doesn't. It's faster for us to test something rather than spend the time it takes to figure out why."

Results do change over time, he adds. For example, they used to lose 15-20% of people who took the first step because they never clicked on the activation email. But when de Albergaria originally tested taking out the activation step, conversions dropped, so they added the step back.

About eight months ago, the drop-off from people who took the first step but never activated rose to an alarming 35%. "We decided to retest the removal of the activation email and this time it worked," he says.

"All of these numbers fluctuate all the time," he notes. "If I look at the conversion rate from the first step to the second step, one day it could be 80% and another day 60%, so you have to do it over time. It's the only way to recognize a valid change."

That's why the team sometimes re-introduces an old concept. "Conditions might have changed," he says, "the audience, the creative used to bring traffic, other circumstances we might not be aware of, so we test something old to see if it might now make a difference."

De Albergaria's team continues to test, though increases have dropped drastically. "We've gone through so many tweaks and changes that we've almost exhausted the opportunities. At the beginning, we increased conversion rates by almost 400%. Now we're happy if we see an increase of 5%."

Useful links related to this article:

Here's your link to 14 creative samples from GoToMyPC's landing page tests, including the current winning control: http://www.marketingsherpa.com/citrix/ad.html