I read with interest a recent Wired Magazine article about Hunch.com – an online decision-support tool that’s similar to Ask.com and Yahoo Answers but with the ability to customize recommendations based on users’ taste profiles and their answers to a set of probing questions related to the topic of interest. The company claims that an average user voluntarily answers 152 questions, allowing Hunch.com to develop intelligent hunches.
Intrigued, I logged on to Hunch.com to see what’s so special about this Web site that propels Internet users to divulge such a huge amount of personal information. Thirty minutes later, against my better judgment, I’d answered 131 seemingly random “Teach Hunch About You” questions, which was a surprisingly therapeutic process. Yes, I can change a lightbulb, and no, I don’t know how to tie a bow tie.
In theory, these questions will enable Hunch.com to understand my personality, tastes, and preferences. That way, when I ask Hunch.com a question later on, it could combine my profile data with my responses to topic-specific quizzes created and edited by other users in order to narrow down to the answers.
Having tried a few questions, what I realize is that for subjective topics like “Where should I move?” and “Should I get married?” Hunch.com’s advice is probably more precise than a Magic 8 Ball or a bad therapist. However, when it comes to product research, like when I asked what car I should buy, Hunch.com’s recommendations were hit-and-miss.
Hunch.com’s predictive power is still a long way from perfection. However, its approach of instantly customizing recommendations by asking customers about their needs and wants is refreshingly different from how e-commerce sites work today.
The common shortfall of such sites is that they only allow users to search for products based on functional attributes. For example, on Amazon.com, laptop computers can be filtered by brand, processor type, display size, hard disk size, and price. It presumes that consumers already have some basic ideas of what solution they need for the job they need to get done.
This is very different from how salespeople make recommendations in retail stores. Imagine yourself walking into an electronics store to buy a laptop. The first thing a good salesperson would ask probably won’t be about CPU or display size. He or she would probe you on your jobs and circumstances: Will you carry it around frequently? Will you use it for video gaming? The salesperson could then deduce from this information which laptop best matches your jobs.
It’s about time that e-commerce sites incorporate this jobs-to-be-done approach in making recommendations to their customers. Using online decision support tools that gather information on customer jobs and circumstances — similar to what Hunch.com has developed — e-commerce sites could help consumers quickly identify the right products and ultimately drive sales. Knowing more about customers’ jobs-to-be-done would allow companies to cross-sell more effectively, giving them the information they need to go beyond a single-product sale to propose a total solution that could help customers achieve their goals. The insight data collected from the decision support tool could also be tied back to purchase and abandoned-cart data, which may shed light on customers’ unmet jobs and innovation opportunities.
To implement an online recommendation tool like Hunch.com’s, e-commerce companies will need to understand customers’ potential jobs and their decision logic (that is, how they make trade-offs between alternatives under different circumstances), and then continually update the program as new products are launched. This is no easy task, but considering the high click-through rate (20%) that Hunch.com has managed to achieve with its online recommendation approach, the investment may well pay for itself.\
Jenny Chung is a senior associate at Innosight.