What does it take to succeed in voice commerce?
Voice assistants are taking the world by storm. The uptake of these devices since their first introduction a few years ago has been unprecedented and continues to grow – by the end of 2018 it is expected that there will be 56.3million voice devices worldwide ( Canalys, 2018).
Naturally, these devices are set to have a significant impact on commerce and will continue to change people’s shopping behaviours. Shopping is becoming an increasingly popular practice on these devices, as they invade homes, often appearing in multiple rooms around the house, as well as in offices, cars and other shared spaces.
With convenience being a crucial factor in eCommerce, it’s only natural that people will turn to these devices to make their purchases. Retailers need to get ahead of the game to ensure they can win in this new domain.
Current usage of Voice Assistants
Research conducted by Kantar & JWT (2017) suggests that the most popular uses of voice assistants are playing music/radio, getting the weather forecast and setting alarms/timers. Purchasing activities always feature lower down these lists. Yet reports suggest that one in five voice assistant users have used them to make purchases, with this figure set to increase over the next year (Walker Sands, 2017) as they become more pervasive in our lives.
It is important that retailers start to develop their strategy towards voice commerce, or they risk missing out on a huge opportunity to connect with their customers.
Essential to develop a "skill"
A “skill” is Amazon’s version of an app. They are used on Amazon’s Echo devices, which are currently the only voice assistants in the UK that facilitate purchasing, so for the purpose of this article, whilst some of the points can and do relate to all voice commerce, the focus will be on devices that have Amazon’s Alexa built in.
Currently there are more than 30,000 skills available globally, in a variety of different categories (food & drink, music & audio, news).
However, there are a lack of skills within retail. Of the top 100 retailers, only one currently has a skill (Ocado, the UK’s fourth largest and most successful eCommerce grocery retailer). This is an area that is progressing rapidly, with increasing amounts of brands developing skills this year, and others to soon follow suit.
So, what makes a useful skill?
Based on current Amazon’s reviews it is hard to gauge which skills are succeeding. Right now there isn’t enough feedback provided for valid outcomes on each of the skills. But it’s worth watching this space to see how the reviewing of skills progresses with more people purchasing devices, using skills and consequently appraising them.
Currently, it seems that functional skills that have high utility appear to work best, such as weather, travel, news, music. This could be due to many reasons, but possibly the simplicity of the language used (discussed further below) and the simplicity of tasks to complete (e.g. Alexa, what’s the weather today?)
Retailers should take on-board these early learnings and consider how they could be applicable when developing their own skills.
AND How to create ONE?
When developing a skill it is important to consider the small window given to capture the customer’s attention. A skill has 0.5 seconds to respond to each request, whether it involves searching, providing information, browsing or purchasing. Skills should be created to work simply and effectively, with minimal effort required from the customer.
When creating a skill, Amazon recommend focussing on conversational design, and getting “people to try it out conversationally”.
Invocation names need to be concise, short and easy to pronounce/understand, in order to work well with users. Often they are the brand or product name (e.g. Alexa, ask Dominos or Alexa, order pizza). Here are some key factors to consider when interacting with customers via voice assistants.
The main limitation across voice assistants appears to be language barriers. Language doesn’t flow as naturally as normal human interaction would, and this can frustrate people. To keep people interested in voice devices, developers need to start thinking of ways to design skills to ensure interactions aren’t too automated. Amazon are developing new features within skills, to facilitate more human-like interactions and to improve exchanges with Echo devices. Right now, English is by far the most developed language on these devices. However with Amazon pushing out Echo devices to 80 different countries at the end of last year, we can expect to see other languages develop and improve across the platforms.
Dictation (the ability to transcribe from voice to text) has progressed greatly over the years - error rates are now less than 5%; on par with humans). However, Natural Language Processing (NLP), that’s the ability to understand the text and respond accordingly, still needs improvement to ensure a seamless experience. Whilst NLP’s progress is hard to quantify, it is evident to anyone conversing with these devices that implement NLP that there is room for improvement. However, with increased use and monetary investment, it is a given that this will happen - and in the near future.
It is only natural that the more or less accurate a voice assistant is in understanding/responding to requests, the more or less people will want to use such devices. So this is something we can expect to see improve in time.
In attempting to make interactions with voice assistants more accurate and human-like, there will be more trust between these devices and users, which will naturally lead to increased use.
Gaining customers' trust
To increase purchasing via voice assistants, trust needs to be built between the user and the device. People don’t fully trust AI yet, for many different reasons, but in relation to purchasing, there are trust issues surrounding selecting the correct product, paying the correct price and with the correct method, selecting the correct delivery method etc.)
Retailers will need to create skills that illicit trust, so people feel comfortable parting with their cash via these means. Ways in which this could be done, could be including variety in responses, particularly opening introduction (instructions don’t need to be read out each time); personalised responses to requests and improvement of language skills.
A resistance to new behaviours
People are willing to adapt their behaviour when using voice assistants, for example, dictating a shipping list or asking for their news ‘flash briefing’. However, they become more reluctant when the behaviour change doesn’t feel natural (e.g. rephrasing a request, or changing the speed in which they talk). A JWT & Mindshare 2017 study showed that half of regular Amazon Alexa users had to change their behaviour to incorporate Alexa into their lives.
Initially, conversing with voice devices can be an uncomfortable experience because it is deemed unnatural to a lot of people. Due to this you need to guide your customer, tell them what to do to engage your skill, this can help to eliminate any confusion, or time wasted trying to figure your skill out.
Give clear instructions as to what the skill can do (e.g. Alexa, ask Ocado what’s in season?, Alexa, ask Ocado if my delivery is on time, Alexa, ask Ocado if I can still edit my order) and tell the customer how to achieve this. In doing this you can ensure your customer is getting full utility from your skill and it will maintain engagement.
Make it easy for your customerS
Most people are still getting to grips with interacting with voice assistants, and so retailers must help their customers as much as they can. They should start by making sure their customers know the correct pronunciation of brand names and products, so they are engaging with the correct skill. Companies should educate people to engage with these devices in the correct way, to make the interaction as effortless as possible.
Go to your customers, don’t wait for them to come to you. Intercepting potential customers at the correct part of their purchasing journey can provide endless opportunities to reach customers (e.g. when they ask for shampoo, when they ask for recommendations on shampoo, when they ask for shampoo for thick hair etc.)
In summary, retailers should act fast to take advantage of the opportunity that voice devices present, before their competitors do so. As Salmon’s Head of Innovation Naji El-Arifi reinforced at Commerce 2020 event “It is not that hard to build a skill on Alexa… Start learning about it because no one is going to tell you the best way to sell to someone using voice.”
Whilst there are still many things to learn about voice commerce, and how people will purchase in the future with these devices, it is certain that purchasing via these devices will rise in popularity. Indeed, Gartner estimates that by 2020, nearly a third of all web browsing will be voice-first and without a screen.
There is still an element of trial and error with utilising these devices to their full extent, however retailers can learn from current available skills, to see what does and doesn’t work. It’s Salmon’s view that it’s important for retailers to make their mark within the voice sector, to avoid being beaten, then locked out, by hungrier competitors. Feedback from our recent Commerce 2020 event highlighted client’s eagerness to learn and move into the realm of voice commerce, which we are keen and ready to facilitate.