Googler Talks Site Testing
E-commerce marketers love to listen to Googlers with advice. Krista Seiden, an analytics advocate at Google, provided plenty of tips on Friday at the 2016 Wharton Customer Analytics Initiative Conference.
She suggested ways marketers can use customer analytics for testing sites in order to adapt and personalize them. During her talk, The Wharton School provided marketers with a live Facebook video feed.
Here are a few takeaways from Seiden’s talk:
Delineate Testing vs. Analysis
Seiden says marketers can test Web page layouts using A/B tools, but analysis is better for content. Analysis allows marketers to grasp the nuances that standard tests don’t, she says.
Decide What to Test
Collect ideas in the organization “and bring everyone along,” Seiden says.
To illustrate, she explained a story about when she used to work on optimization for Google clients and she, too, primarily emailed the sales and marketing teams about what tests they’d like to see each quarter. Then she realized other teams had great insights and started including, for instance, developers on the emails.
‘Remember to Think About the Low-Hanging Fruit’
“Very, very simple tests can have a big impact,” she says.
For instance, testing the call to action “start free trial” vs. “get started” yielded great results, Seiden recalls. “Get started” blew the previous CTA away, with a 21 percent increase in conversions.
Gather key stakeholders, prioritize the ideas and be transparent with the organization by making the information available throughout the company, she says.
Know What You’re Testing
Multivariate testing is great, as long as marketers know what the results mean, Seiden says. Any test where marketers can’t see what drove the results is a bad test, she says.
Use the Data to Personalize Web Experiences
Tailor customer experiences on sites based on geography, traffic sources and actions taken on the marketers’ sites.
For instance, the Japanese version of the Google Apps for Business website in 2013 needed to test a guided flow of every step of the trial process, rather than pushing Japanese businesspeople into calling Google sales. (Seiden says while the B-to-B customers there are very process-driven, contacting sales would be out of the question for them that early in the cycle.)
“This led to a 693 percent increase in clicks on this button,” she says of the one Google switched from “contact sales” to a guided flow, “which is huge, and an 8 percent increase in people actually starting the trial process. In Japan historically, we had a really low trial process start rate online, because they like this handholding.”
But the guided flow flopped in the U.S. and the U.K., because the guided flow was very much a local solution for the Japanese B-to-B audience, she says.
So Googlers decided to redesign the entire site experience for the Japanese audience.
Understanding referral sources and personalizing experiences based on that information can be as easy as making the language on the landing page or app match the language from the referral source, she says.
Understanding customers at the action level can be simple, too, Seiden says. For instance, if she visits an outdoor gear site and goes to an Adirondack chair product page, then flips back to the home page, the home page can say something clever like, “Our website is busy. Take a seat.” That’s an example of a test, she says.
What do you think, marketers?
Please respond in the comments section below.
Related story: 6 Steps to Actionable Analytics