When the Echo launched back in 2014, it was assumed that it would be more than just an assistant, that it would become Amazon’s new store front: “its driving force, the beat in its heart, will be to accept your money as efficiently as possible.” As recently as this March, a a market research report predicted that “voice or conversation commerce [will] become a $40 billion market by 2022” up from just $2 billion today.
Yet almost four years later, that doesn’t seem to be the case. Last week the The Information, citing an internal Amazon stats, reported that “only about 2% of the people with devices that use Amazon’s Alexa intelligent assistant—mostly Amazon’s own Echo line of speakers—have made a purchase with their voices so far in 2018″ and of that did try voice shopping, “about 90% didn’t try it again.” Even with the about 50 million devices that Amazon has sold, that’s a pretty low number, with an even lower number of repeaters.
It’s interesting to explore why.
One well-adopted feature that was part of Echo from the launch is “add to shopping list,” which also exists on Google Home. Around 36% of all voice assistant users (on both mobile and smart speakers) use the shopping list feature. Lists are a a natural use of voice assistant because they’re intuitive and simple: just “add salt to my shopping list.” Speaker location can also promote the use of shopping lists, especially when located in the kitchen, where 41.4% of US smart speakers are located, and the living room, where 45.9% of speakers are placed. It makes sense to add an item to the shopping list when opening the fridge and realizing it’s missing.
So if a list already exists and is being used, why aren’t more users placing orders?
First, it might be because it’s a complicated path from a generic item like “laundry detergent” to the voice equivalent of a product page, with a brand, size, and price. On Amazon a search for “laundry detergent” brings up over 4,000 results, which drop down to over 1,000 results when filtered for Prime. Even a specific brand will return multiple results. Cinnamon Toast Crunch cereal has 122 products on Amazon, which drop down to 67 when filtered for Prime. Once reaching a product page, Amazon offers multiple sellers at multiple price points and shipping costs that add another layer of complexity. With that amount of choice, ordering a product requires multiple questions, which even dedicated voice shoppers might despair of doing.
Second, many purchases might never fit into voice driven shopping because they require visuals before decisions are made. These include clothes, prints, shoes, bags, linens, etc. I can buy laundry detergent sight unseen but not a t-shirt.
Finally, sometimes consumers require more information about first time purchases, such as ratings and reviews, especially for larger purchases and electronics.
These three limitations might impede first time purchases but a first-time voice driven purchase on Echo is not necessarily a first-time purchase on Amazon. Again, I have no insight into Amazon’s products or data but a customer’s purchase history can be a shortcut when trying to convert a shopping list item to an order. When a user says “buy laundry detergent” the result shouldn’t be all 4,000 results but rather the laundry detergent bought last month. Repeat purchases of dry groceries, toiletries, household sundries, and office supplies should be a no-brainer.
This is why it’s interesting that Amazon hasn’t made this work. With the logistics of Prime and one-click ordering already set up, why don’t existing Amazon customers shop on Echo?
One possibility is that in trying to remove friction from the purchase, Amazon has created uncertainty. Usually, removing friction from a transaction involves hiding information and skipping decision making points. Perhaps in removing all detailed information about a purchase, Amazon has left customers uncertain?
Another cause could be pricing changes. Amazon prides itself on dynamic pricing, raising prices to match increased demand to maximize revenue, and shipping costs can vary depending on what seller fulfills the order. Users might be OK with paying a price when they see it before they place their order on the web or app, as opposed to being surprised by a higher price when their card is charge after a voice order. Are voice customers wary of Amazon’s price changes? Would predictiability increase trust and drive orders? The low rate of repeat purchases on Amazon seems to signify that that trust in pricing is lacking.
Finally, perhaps the entire flow is wrong. A few months ago I wrote about cooking with voice and how the format needs to change so that users can actually follow a recipe. Simply reading out loud a recipe as it’s written, ingredients and amounts first, instructions second, makes it difficult to follow. What works in another format, doesn’t work for voice. Shopping also has to change for customers to adopt it, perhaps offering better options, learned from previous purchases, perhaps by smartly reducing options, making choice easier, or maybe being more predictable with less pricing changes, or maybe just presenting them at a different stage of the process. Either way, I’ll be really surprised if shopping via voice doesn’t take off soon. It’s just too convenient.