Stuck on Website ‘Help Island’? First UW iSchool Spinout Has the Answer
[Corrected 5/23/14, 10:53 am. See below.] The University of Washington Information School has more than a century of history, and now, it’s first startup company, AnswerDash, which aims to improve self-service online help for e-commerce, government, and other Websites.
While startup companies and technologies with commercial potential regularly emerge from the larger UW Computer Science and Engineering departments, that hasn’t been the case with the iSchool, which fosters technology innovation in areas such as human-computer interaction and information management and science.
Linden Rhoads, vice provost of the UW Center for Commercialization (C4C), thinks AnswerDash could be the beginning of a trend.
People familiar with UW Computer Science and Engineering “don’t realize the extent to which we have this world-famous iSchool that looks at similar but different specialized subject matter,” she says. The iSchool ranked third among library and information studies graduate programs and fourth for information systems programs by U.S. News & World Report.
Within the iSchool, which began as the Department of Library Economy in 1911, AnswerDash is being celebrated as a milestone. Professor and iSchool dean Harry Bruce says the company is indicative of the kind of research done by faculty and students nowadays on facilitating and deepening people’s experience with information.
“As the iSchool grows”—it hired eight new faculty this year for a total of 49, including lecturers—”and we continue to attract highly creative PhD students, the potential for further commercialization of our research will undoubtedly increase,” Bruce says in an e-mail. “We are excited by the prospect of deepening our impact through the development of successful companies that provide innovative tools, systems, and services that make information work.” [An earlier version suggested the total iSchool faculty was 42, including the new hires. There are currently 42 faculty, with the new hires growing the ranks to 49 to start next year.]
AnswerDash—which began last year under the name Qazzow and just changed its name—is based on a four-year research effort by then-iSchool PhD student Parmit Chilana (now a professor at University of Waterloo) and her co-founders, Jake Wobbrock and Andy Ko, both professors at UW.
The idea is to make finding online help easier and faster for consumers, and less onerous for businesses to provide it.
“Web self-service is an important area because it puts the burden on the user and takes it off the company in terms of cost,” says Wobbrock, the company’s president and CEO (pictured, at left above, with Ko and Chilana). “At the same time, humans want to self-serve because they feel self-sufficient and confident.”
There’s real money at stake, too. Shoppers are more likely to abandon their transaction if they can’t find the answer they need, he says. AnswerDash estimates this is the cause of upwards of $8 billion in lost e-commerce revenue in the U.S. each year.
There are problems with most of the standard online help scenarios today, Wobbrock says. So called “help islands”—separate pages populated by things like knowledge base articles, isolated lists of frequently asked questions, and endless user forums—are typically cumbersome and require lots of digging to find what is usually a one-sentence answer to a simple question. They are used infrequently and often fail to provide an answer, Wobbrock says. Live online chats can be more effective, but take time and company resources. Help is even harder to find on mobile Web pages and apps, which account for a fast-growing portion of retail e-commerce sales.
“Mobile self-service almost doesn’t exist. So we give them an F-minus, which is also a grade that doesn’t exist,” Wobbrock says.
AnswerDash’s answer is software as a service that offers contextual help. It combines patent-pending object-oriented search, a bit of crowd-sourcing, and the 80-20 rule (the most common questions are few in number but asked most frequently) to provide what looks to me like a significantly improved online help experience.
The most common questions and answers accumulate over time, building a corpus of easy-to-access information that will take care of most customers’ needs, freeing up a company’s customer support teams for the trickier questions.
“There will always be a long-tail of questions that we don’t intend to handle, and those are appropriately escalated to customer support,” Wobbrock says.
Object-oriented search seems to be the real secret sauce here, making the process fast and easy for customers with questions. Clicking on the image of the mouse triggers a search of all the questions asked as well as answers given on that site, with the most relevant ones—based on things like how recently and frequently they’re asked—presented first. Contrast that to a typical search through help island, which starts with typing and then skimming through results to select one that might contain an answer to your question.
“So it flips the formulation and selection process around,” says Wobbrock, whose expertise is in human-computer interaction. “And as a result, we don’t get duplicates. We have about 10,000 questions in our system and never had a duplicate because people first look and say, ‘Oh, that’s my question,’ even if it’s maybe not how they would have typed it.”
That’s one of the efficiency benefits of this contextual, object-oriented approach. It addresses the so-called vocabulary problem in information technology: “people use a surprisingly great variety of words to refer to the same thing,” as the authors of a widely-cited paper published in Communications of the Association for Computing Machinery put it back in 1987 (PDF).
“You say, ‘How do I insert an image?’ I say, ‘How do I upload a photo?’ He says, ‘How do I add a picture?’” explains Wobbrock. “All different nouns and verbs, but we mean the same thing.”
Because consumers aren’t stuck typing the same question over and over in slightly different ways, customer support teams don’t have to waste time answer it repeatedly.
AnswerDash also allows people to ask questions using far fewer words, part of why it’s particularly well-suited to mobile apps and Websites. Your question about the mouse could be as simple as “Will this go on sale?” The people answering the question know what this you’re referring to because they can see precisely where you asked it on their Web site, providing the necessary context. (This contextual requirement is something linguists call deixis.)
“That’s a powerful thing that no one has really taken advantage of on the Web yet,” Wobbrock says.
It also points to another reason businesses would buy the AnswerDash service, which is sold on a tiered subscription model based on how many times users click on the Q&A tab and starting at $24 a month.
“When a user clicks on a question and sees an answer, that informs a whole set of analytics we can provide the company,” Wobbrock says.
Not only does AnswerDash reduce abandoned transactions by allowing quick answers to customer questions, it also shows companies fine-grained details about which specific questions and answers lead to sales. For software-as-a-service companies—an industry AnswerDash is targeting initially—this is important for reducing customer churn, particularly in the early phases of adoption when new users are likely to have more questions.
AnswerDash received a $500,000 seed investment from the W Fund last fall. In December, that was folded into a $2.4 million Series A round led by Voyager Capital and WRF Capital, with participation from Summit Capital, Geoff Entress and four other angel investors. It just announced new directors and advisors, including former Drugstore.com CEO Dawn Lepore and former Decide.com CEO Mike Fridgen. Other board members include Voyager managing director Bill McAleer and Seattle technology executive Ken Myer. [An earlier version incorrectly suggested Fridgen was with Voyager and omitted the reference to other board members.]
The eight-person company housed at the UW C4C has several beta customers including Moz, Big Fish Games, and the CIO of the state of Washington, Wobbrock says, adding that the product will come out of beta soon.
Wobbrock and Ko were granted leave to work on the startup, which underscores the support they feel from the iSchool, Wobbrock says.
“It’s a significant impact on the department to have two of its faculty not really in the trenches,” he says.
That support also points to an expanding definition of academic success, traditionally measured by scholarly journal articles and federal grant funding.
“The Information School realizes that by spinning things out, that’s another form of impact, that’s another way of impacting the world,” Wobbrock says.