Little tweaks pay big for Skinner site

If there are two constants in direct marketing that have translated to the Internet, they are that people never do what we think they’ll do, and that little things can make a huge difference in response.

Skinner Auctioneers and Appraisers of Fine Art recently found that by testing variations of the layout, wording, and placement of its e-mail newsletter sign-up form, it was able to increase registrations by hundreds of percentage points, and well over 1,000% in the case of repeat visitors. And that was just one tweak that paid off handsomely for the Boston-based auction house.

As is the case with most businesses, Skinner’s e-commerce site is playing an ever increasing role in the auctioneer’s sales. But just how much of a role is hard to tell because there is evidence that the site is driving offline bidding, says vice president Kerry Shrives.

“We certainly know that sales from our printed catalog have been stagnant because the content is free and available on our Website,” says Shrives. “People who come to our auctions often show up with all kinds of printouts that they’ve generated. We know that people are looking to the Web for content.”

The Internet is also creating smarter bidders, she says.

“People can get a head start doing their homework to get an idea of what they may be interested in pursuing, which gives them an advantage when they hit the floor,” says Shrives. “If people are doing their research ahead of time, that often can lead to higher bids because they come at it with more information.”

To help determine Website design changes that would boost performance, Skinner hired multivariate testing firm SiteSpect.

Among Skinner’s goals was to increase subscribers to its e-mail newsletter and wean as many as possible off its print newsletter, says SiteSpect president Eric Hansen. “They were getting a trickle of new subscribers every week, but nothing substantial,” he says. “We ran a series of tests, and with each test we were able to improve the rate at which people were subscribing.”

With an average of some 20 or so page views per visit, activity at SkinnerInc.com indicates a seriously loyal following that should be an easy target for a free e-mail newsletter, says Hansen. The trick was to find out why they weren’t subscribing in the numbers Skinner executives would have liked, he adds.

“The first question was: ‘Why wouldn’t they subscribe?’” says Hansen. “It’s free and doesn’t ask for any information. So the first test was to see if maybe something about the signup box itself wasn’t right.”

SiteSpect tried a variety of different calls to action in the signup box, including one that said: “We won’t spam you,” says Hansen. The firm also tested different colors, fonts, and text sizes.

For example, SiteSpect tried a gray background with red text, a white background with black text and a red background with white text.

The winner: a signup box with a red background and white, bold, sans serif type with a no-spam disclaimer, says Hansen. Direct marketing creatives will instantly recognize that the winning box goes against DM conventional wisdom that says reverse, sans serif type is generally a design no-no. But there is no arguing with the results of a well-crafted test.

As for the signup box using the more conventional white background with black text: “That one underperformed,” says Hansen.

Skinner’s original e-mail signup box featured a dark gray background with white text that said: “Subscribe to Skinner’s newsletter for alerts about upcoming auctions,” and a submit button that said: “go.”

After a series of tests on the language, SiteSpect determined the winner to be: “Subscribe to Skinner’s Ezine to receive e-mails about our upcoming auctions and events” and a submit button that said “subscribe.”

The winning red-and-white signup box with the new language is outperforming the original by 210%, says Hansen. “But this wasn’t the blockbuster test,” he says. “We did a series of subsequent tests because the overall subscription rate still wasn’t super high.”

A surge in signups

After a series of tests, SiteSpect found that by putting e-mail signup boxes on so-called department pages where items are grouped by category, the firm was able to increase e-mail signups by 590% among new visitors, and by 1,750% among repeat visitors.

Why did the department page signup boxes boost registrations so significantly? “Probably what happened was they never saw the sign-up box in the first place because they spent only a cursory amount of time on the home page,” Hansen says. “They’re not interested in auction houses; they’re interested in Asian works of art or they’re interested in antique musical instruments and that’s where they spend all their time.”

Another reason shoppers may have been missing the home page signup box is that people often don’t enter sites through home pages, says Hansen. “They could have bookmarked a department page,” he says.

Interestingly, the red box with white type that outperformed all the other boxes on the home page was not the winner on SkinnerInc.com‘s department pages. For whatever reason, visitors to the department pages responded most to a gray box with black type.

One lesson to be learned from the different test results on the department pages, Hansen notes, is that a site operator should never assume that a technique that works on the home page can simply be rolled out to the rest of the site without testing.

“And that’s just within one site,” he says. “If you apply that [benchmark] to another site, your potential margin of difference is even greater. What works on Skinner Auctions may not work on Sotheby’s at all, or vice versa. The good news is it really wasn’t that hard to run these tests; it didn’t cost that much and the results were substantial.”

Adds Shrives: “We’ve had a couple of things that have been successful and totally debunked what we thought was a given. How a customer uses a site isn’t necessarily how the site designer would expect.”

She says Skinner currently has more than 25,000 subscribers to its e-mail newsletter. The auctioneer uses e-mail service provider Constant Contact to send its newsletter.

When asked about the quality of the e-mail addresses Skinner is getting, Shrives says the nature of the art-and-antique auction business makes it difficult to tell.

“If you’re looking for a dining table, once you buy it, you’ve bought your table,” she says. “You might be looking for something specific and may not find it for a long time, so at this point we’re happy to increase the spread of our potential bidders and keep them informed about what we’re doing.”

She adds that Skinner’s customer demographics vary widely. “It’s a real mix,” she says. “There are people who are collectors who are looking to buy things and keep them for themselves; there are dealers who purchase for resale; there are institutions that are collecting.”

Skinner executives believe increasing customer engagement in the Website is crucial.

“We’re constantly looking for ways to get people to spend more time on the site, look at more content, and hopefully to register and to bid,” says Shrives.

Bigger is better

In another test, Shrives and SiteSpect found that by increasing the size of product images, they could significantly increase bidding activity on the site.

“If you looked at it real carefully, it wasn’t any more detailed, it was grainier,” says Hansen. “It was just a bigger image.”

But by simply increasing product image sizes from 250 pixels to 350 pixels, Skinner achieved a 429% boost in conversion-to-bid rates, says Hansen. “Just making the image larger persuaded people to bid much more frequently,” he adds.

Shrives adds that the improved performance generated by enlarging images gave her the idea to increase the size of the thumbnails on the site, as well. But increasing the thumbnail sizes was an utter failure, says Shrives. “It seemed to be a logical step, but it was quite the opposite,” she says. “It caused a decrease in items people looked at.”

Though Shrives’ background is in the arts, she is beginning to sound an awful lot like a direct marketer.

“We test, and if the results are statistically significant we do it,” she says. “This has been a very good process for us.”