Conversational Search for Increasing Ecommerce Revenue
Conversational Search is a product concept for increasing ecommerce revenue by making shopping experiences
more efficient and more satisfying for consumers. By modeling product search and ecommerce interfaces as special
cases of human conversation, Conversational Search leads to a development plan with concrete, incremental steps
for embedding experiences into online search sites and other collaboration/communication products. My goal is to
use Conversational Search in ecommerce as a starting point for changing the landscape of all online search experiences.
Conversational Search is not about natural language processing or speech interfaces. It is a development framework
that breaks the complex problem of "improving the ecommerce experience" into smaller, tractable parts that can be
individually solved with new types of content presentation and new user controls. Each aspect of successful
ecommerce conversations can be modeled, prototyped in software, tested, and refined, all targeted at the current
client-side delivery platform. Interface improvements for individual aspects of ecommerce conversations can be
embedded into any existing search site in the near-term. Then a unified, seamless, and revolutionary experience can be developed in
Today's commodity search experiences offer 'keywords in/listings out', a forced interaction pattern that is limiting,
inefficient, and often frustrating. Search interfaces do little to help the user to converge on the right piece(s) of
content for a particular intent, yet that is the universal goal of searching.
Conversational Search is based on the core truth that human-to-human conversations—the give-and-take of open-
ended exchanges that lead to new viewpoints or new agreements—are the most powerful and efficient process for
clarifying human intents. In conversational software interfaces, the service presents each piece of content based on
fine-gained awareness of what the user knows and how the listener prefers to learn. As a result, the user absorbs new
content faster and with less cognitive effort. Conversely, when reacting to or directing the service, the user has more
fine-grained control and can focus the content flow by natural functions such as a “yes-and-no button”—a single
command that allows the user to say how the content is ‘on point’ but also how it is not quite right. Overall,
conversational interfaces bring human-computer interaction closer to the power and naturalness of everyday
The roadmap of Conversational Search begins by examining the evolving cognitive model of participants in
successful person-to-person conversations in everyday life. These models become the foundation for software
interfaces that directly and efficiently support such conversations. For ecommerce conversations to be successful,
the user must have a combination of rational understanding and emotional confidence in order to make a purchase
decision. Rational understanding requires developing a shared language with the user; enabling the user to learn
about possibilities; and expediting the user and the online service to reaching agreement on specific choices for the
purchase. Emotional confidence also has three aspects: confidence the product will fulfill needs and desires;
confidence in the choice of a specific product; and confidence in a particular merchant at a particular price.
Beyond search and ecommerce, a range of products and services can also be significantly improved. "Vertical sites"
that specialize in health information, local businesses, or end-user technical support have huge potential for
improved customer satisfaction, retention, and revenue. Similarly, enterprise collaboration platforms demand
efficient, confident conversations with trackable agreements as outcomes.
There is precedent for building software applications from effective models of conversation. Experimental work
from the 1950s through the 1970s led to a widely recognized and repeatable measure of conversational agreement
and knowledge acquisition, now an accepted part of the literature of educational psychology.