A startup called AnswerDash is launching a self-help service today that brings contextual help down to a single word or image.
“AnswerDash is about giving answers in context when [visitors] use a website or web app,” CEO and co-founder Jake Wobbrock told VentureBeat. By contrast, Wobbrock said, typical website help sections, FAQs, knowledge bases, and site-wide search engines are standalone “help islands.”
Let’s say you’re reading a website on a computer or mobile device, and you have a question about something mentioned or shown in the middle of a page.
Normally, a user might enter a query phrase into a general or help section search box. The results, once reviewed, may or may not relate to the question at hand. Or they might go to the site’s FAQ or a help community.
Instead, AnswerDash offers a small tab on the right side of each page. When clicked or touched, the page becomes grayed out. Words, images, video, or any other user interface elements then become individual hotspots.
Hovering over a hotspot for, say, the phrase “hospital costs” opens a small window with the most popular questions relating to that phrase, as well as their answers. There are also options in the window for asking a new question or for getting live help via chat or phone.
New questions are emailed to the site, with replies emailed back to the user and automatically entered into the AnswerDash database for the next time that hotspot or a similar one is chosen. The company recommends that each site prepopulate only about 20 questions and answers, so that users’ questions — what visitors actually want to know — constitute most of the inventory.
Natural Language Engine
Within a few days of usage, AnswerDash says, the most common inquiries will be covered by the growing inventory of new questions and answers. A site can decide if it wants to edit the new Q&As before they are added.
Metadata, contextual information, and previous questions are employed to provide a context for relating questions to image- or video-based hotspots. A natural language engine, customized by AnswerDash from an open source version, parses the page’s text or metadata in real-time to find the questions and answers relevant to a given word/image hotspot. Answers can be provided by video, links, or images, as well as text.
The company, initiated in the fall of 2012 and incorporated last winter, is the result of National Science Foundation-funded research conducted in two human interface labs at the University of Washington in Seattle. More specifically, it was the subject of a dissertation by one of the three co-founders, then-student Parmit Chilana, who is now an assistant professor at the University of Waterloo.
The company’s research showed, for instance, that more than 95 percent of questions by website visitors “has been triggered by something seen on the site,” Wobbrock said. About 80 percent of all visitor questions, the researchers found, can be covered from the inventory of questions already answered in a site that has regularly used AnswerDash. The other 20 percent are directed to human-based help, via email, phone, or a separately provided but integrated live chat.
Wobbrock, also a professor in the Information School at Washington, told us that a previous beta solution has been tested with a few selected companies. Late last year, the company raised $2.54 million from Seattle-area angel and VC investors.
While there are dozens of companies that provide some form of help systems or communities, Wobbrock said none offer “this kind of point-and-click contextual help” at this granularity. A few — he pointed to UserVoice and Qualaroo — offer broader answers via what he called “text box widgets,” which provides search results synched for a page. Others, such as a new “contextual help” offering introduced by Zendesk last month for admins, are keyed to menus.
A subscription model, starting at $79 per month, is based on how much the site’s visitors actually use the service. Initial customers include the US Green Building Council, PetHub, and Red Awning.