Personal digital assistants are taking the world by storm. With Siri and Google Assistant on our phones and Amazon Echo and Google Home in our homes, we are on the cusp of a world where “voice-based” products and experiences are ubiquitous.
These are still early days, and the primary uses for voice interfaces appear to be playing music, setting timers, and controlling smart homes. Voice search in e-commerce is largely used for re-ordering consumables, where people don’t need (or want) to make a lot of choices.
Many industry insiders believe that while voice search clearly has a place in the e-commerce landscape, it will never be used for “serious” shopping. Just last month, Charlie Cole, Samsonite’s Global Chief eCommerce Officer, told the crowd at NRF that “No one is going to say ‘Hey Google/Alexa, order me a $1,200 cashmere sweater.’”
Whether today’s low rate of “conversational commerce” reflects a temporary state of affairs, or an intrinsic technological limitation, remains to be seen.
But if history has taught us anything about predicting the impact of technology on consumer behavior it is this: never say never. So often, the thing which people think will never happen invariably does, and the naysayers are left dealing with bigger problems than they ever thought possible.
In 1966, Time Magazine published an essay titled “The Futurists,” which imagined what the world would be like in the year 2000. The essay predicted that while remote shopping would be possible, it would never take off because “women like to get out of the house, like to handle the merchandise, like to be able to change their minds.” That may be true, but it hasn’t stopped men and women from boosting e-commerce activity year over year.
The consumer has spoken, literally. Shoppers clearly recognize the value of voice commerce, and want it to work in more ways. So what will it take for consumers to use voice search to find and buy that $1200 cashmere sweater?
Consumers Expect Understanding
For 25 years, online shoppers have felt misunderstood by the search engines they use to find products. We’ve learned to look for “red dress long” rather than “a long red dress” — and by now, we know better than to search for “a long red dress with a button-up back for under $300”.
But today voice interfaces are encouraging consumers to speak more naturally. People speak about three times faster than they type, and we don’t naturally limit our expressions to keywords.
Voice search returns freedom of expression to the consumer and shifts the responsibility of understanding to technology. Consumers expect machines to understand what they say — and what they mean.
But understanding long, natural-language queries is a lot harder than handling short keyword searches. If a shopper asks, “Show me a lightweight city bike,” the machine has to determine that a “city bike” is a type of bike, that the searcher’s intent is to find one available for purchase, and that s/he wants one that is “lightweight” — which is itself a subjective term.
Conversation Should Clarify Intent
Even with a fairly specific request, there’s often a wide range of products that could satisfy it. It’s also unclear how to arrange the results by order of relevance without additional input from the consumer. In the above example, a human sales assistant would probably follow up with a clarifying question like “Who is the bike for?” or “What is your budget?”. People clarify intent through intelligent conversation.
Intelligent conversation is as much about asking the right questions as it is about giving the right answers.
On one hand, “Are you interested in bicycles?” would be a silly follow-up question, since the answer should be obvious. On the other hand, “Are you interested in a titanium bicycle?” is too specific and exotic: the answer is probably no and doesn’t meaningfully advance the consumer’s progress towards a purchase. The machine has to ask the right questions — and ideally as few as possible.
Still, it’s important to ask enough questions. Without knowing a bit more about the consumer’s needs (e.g., his or her budget), there’s no way for the machine to narrow down the selection of results. And listening to a sequence of options is far more painful than scrolling through a page of search results. Conversational commerce hinges on clear intent.
How will we reach the point where we use voice interfaces to make major purchases, like a $1200 cashmere sweater? The key is trust.
Machines must consistently demonstrate that they understand what consumers are asking for. They need to do so by responding with relevant results and intelligent questions that clarify the matter. Trust grows from consistency over time — which means that it will take a long time to develop that trust but a short time to lose it. It’s a high bar.
Still, the success of e-commerce proves we can get there. E-commerce search today is hardly perfect, but it’s earned enough consumer trust that it now accounts for about 10% of retail sales. That growing share reflects consumers’ willingness to trust an online shopping experience and benefit from its convenience, despite their inability to “handle the merchandise”.
Most of us are more comfortable speaking than typing. Once machines convince us that they can understand what we are asking for, we’ll happily throw away our keyboards. We’ll opt-in to voice search.
It’s only a matter of time that people discover the immense convenience, speed and ease conversational commerce offers and start to rely on it for online shopping. For online retailers who can get a head start and invest in the right technologies – this is an incredible opportunity to grab with both hands.