Critical New UX Design Optimization Research

OLC attended New York Technology Council's event at Anchin: Critical New UX Design Optimization Research featuring Paul W. Thurman. Thurman teaches strategic management and data analysis courses at SIPA and at Columbia University. He is a highly regarded customer experience optimization specialist. His professional experience includes Booze Allen Hamilton and American Express. Thurman specializes parsing complex consumer behavior for business optimization and innovation.

"User experience starts with you," Paul Thurman said. "As a designer, I don't care what you do, but I do care how you do it." He said that figuring out how the user comes to their decision is the toughest part. "This is the new frontier," he said, "figuring out your thought process." Thurman explained that the brain is able to process images 60,000 times faster than words, primarily because "you have to read the words. Why do you think your kids figured out the iPad before you did?"

Thurman went to explain that there is a big challenge in trying to speak through pictures. "Figuring out what picture is easy, but how you got there is tough." He revealed that to be beautiful means to be usable. "If it looks nice, it likely makes an impression. Visual grabs much faster than text and although beauty may be in the eye of the beholder, it is not blind. It actually matters—beautiful means it gets adopted quickly and earlier than other products that aren't," he said.

"Sample size, does it matter?" he asked. "There are usually thousands of volunteers, paid or unpaid, that test the usability of products everyday. This is how products are generally tested. It turns out that about 90 percent of data is repetitive and useless. You can get the exact same data with just a handful of people and get good data." Thurman gave a quick example of clinical trials done by the FDA, where the organization tests just 5-600 people before releasing it into the public. "Phase 1 is tested on about 20 people to see when side effects occur. Phase 2 is the observation of side effects, which is about 150 people. And phase 3 is efficacy, which is tested on 2-300 people." Thurman emphasized that you don't need a big set of data.

He also focused on quality data over the quantity of data. "Statistical theory is old," Thurman said. "Figuring out interface design is new. With a bigger n, you get bigger noise in products. More options don’t mean it's a better choice. Few choices mean more people find the features more valuable." He presented present-day society has "big data" but with "little information," less "knowledge" and even less "wisdom." Contemporary society, according to Thurman is "data rich but information poor."

"Over 72 percent of the functionality in Microsoft Word goes unused," Thurman said. "Lots of software we tout as great, we hardly ever use its functions."

He presented his first trend: Segmentation. "Don't treat every customer the same," he said. "Think big, but sample small. Think in segments and not the population. The wisdom of crowds is a myth. If you listen to people to create your website, you're going to end up with an average website. Don't cater to a population, work your website to fit the individual. Spotify comes close to doing this by analyzing your music preferences and changing the recommendations around to fit your taste." According to Thurman, the web is lagging behind in segmentation even though it is free.

"There's a problem that we wish that we all had," Thurman said. "What makes an interface so compelling that we just can't stop using it?" He introduced Angry Birds as an example. He broke down the layers into simplicity, timing (short-term memory loss), mystery, music and the unmeasurables (plush toys).

Thurman introduced the second trend, which revolved around simplification and beauty. "We tend to design things for everybody," he said. "We should attract then treat, attract people you want and then change the user interface as they are watching or refreshing the page. This way, it keeps the mystery and maintains the short-term memory aspect."

"Now, do you sell books or maintain a library?" he asked. "Libraries are big structures with free use of content, where other places would charge you a fee, like supermarkets, health clubs or parking garages. A national consumer electronic chain revealed that their store visits are up but their profits are down. They have incentives in place, but people aren't buying things. Instead, they're sucking the information out from the sales reps and then going price shopping. The salespeople are becoming librarians," he said.

"The abandon rate in ecommerce in 2011/2012 is around two-thirds, that is, two-thirds of all shopping carts are abandoned. People do this if delivery dates that are seven or more days, 38 percent abandon their carts. They do it when they see the tax, shipping and handling charges or they do it for the pure shopping effects like bundling or checking to see if they are going to get a deal," Thurman said.

Regarding A/B testing, only 44 percent of major online sellers use split-testing software. "For the past two years, simple A/B tests, based on old statistical methods, have resulted in most conversion improvements in ecommerce." Thurman quoted Einstein to emphasize that doing the same thing over and over again expecting a different outcome as insane. "Use more controversial tests," he said. "Use different tests to see more surprising and insightful results. Don't test just the product difference, but test the customer differences too. Keep the tests simple. If you think PhD statisticians are necessary to do it, you're probably over-complicating things."

"Are you serving customers or managing servers?" Thurman asked. "In the IT business, most databases are built by people in databases, not by people who are in the solution-selling business. The vast majority of databases optimized for databases and network performance are built by database managers, so there's no information to work with for people in the solution-selling side. Barely 10 percent of major ecommerce platforms are built for customer data analysis."

Thurman introduced the fourth trend, retention over acquisition. "Keep interfaces fresh and new," he said. "You get 105 percent more conversion lift with using ratings and reviews—just look at Amazon and how much time they spent on their review system. Reward loyalty, but in the right way. Look for new ways for users to pay for the business—currency in many ways. Invest in the customer lifetime value, increase focus on the lifetime element as an independent value."

Thurman asked if the audience, when launching a new product, tests it and learns from it to iterate or launches it and prays that it works. He said that testing and learning and iterating the product is the best way (also used in science experiments). The best entrepreneurs failed multiple times before succeeding. "The number of marketing and promotional failure is highly correlated with customer profitability," he said. "How often you don't get it right connects with success." Thurman also recommended that the audience talk to customers. "Talk to even non-customers, ask them why they aren't buying from you, or why they're leaving, or why they haven't bought from you yet." He revealed that collaborating with other people outside of your network is also critical because it ends up with efficacy and saves valuable time.

He outlined the progress of screens throughout the advancement of technology—TV (head-back device) to PC (head-forward device) to mobile. "Ninety percent of information we get daily are from screens," Thurman said. "In the old days, we used to sit in front of a screen, but now screens are just as mobile as we are. Mobile is the new frontier for data analytics." In mobile, the developers are now releasing first to mobile to test and iterate and launching web apps after. Thurman said that new science us needed in a multi-screen world. "Measuring how you behave or react when you are not stationary—due to environmental covariates—is tough. We need new science to understand more split-second decision parameters and to understand mobile user behavior or data capture because it is even more limited than web-based nets."

Thurman said to think about what users need, not what developers want for themselves and to ask who really is the customer. 

NYTC's Legal Essentials for Startups - Part 2 of 3: