Emma Herrod investigates how voice and visual are changing search and whether they hail the future death of the keyboard.
From the first home PCs, to the Blackberry and Apple’s series of iPhones we have become familiar with the keyboard. Regardless of how many fingers you use on the keyboard or whether you can talk, read and type simultaneously, there is one thing that we all do – use the keyboard to search online.
If we want to find the answer to a question, buy something or just browse, search has become second nature, and it’s not that long ago that the keyboard was the main way to interact with a search engine.
Even Google Suggest, which autocompletes searches as we type, has made entering searches faster and meant that we don’t even have to worry about spelling. Jump forward to 2016 and 20% of all searches in the Google app were spoken and not typed. A year later, almost 70% of queries to Google Assistant on any device were made in natural language – that is, people speaking in full sentences as opposed to just using the key words as they would when typing into a search box, explains Alessandra Alari, Head of Search and Digital User Experiences, Google UK.
It’s not surprising then to discover that consumer adoption of voice technology has been faster than any product since the smartphone and that ownership of smart speakers in the UK doubled from 5% to 10% between Q3 2017 and Q1 2018, according to YouGov. It found that 10% of homes own a device. The majority of people who own a voice-activated speaker say that it feels natural speaking to it.
Morrisons, Argos, Tesco and Ocado were amongst the first retailers in the UK to launch voice-activated shopping services to match the popularity of smart speaker ownership and the rise of voice searches on smartphones.
Morrisons, for example, launched an Alexa Skill at the end of 2017 so that customers could ask Alexa to add items to their shopping basket as they ran out. The app then suggests the right item based on previous purchases. Shoppers can also ask Alexa to edit their order, ask if their delivery is on time or check their basket to see if an item has been requested.
The Argos Voice Shop service takes things a step further though since it uses Google Assistant which can be accessed from any device such as a smartphone and not just a smart speaker.
A shopper can search for a product, check availability or find their nearest store simply by activating Google Assistant by saying “OK Google, talk to Argos” followed by the item they are searching for such as a kettle, or say “OK Google, ask Argos to find me a kettle”
This opens a dialogue with Argos’ Voice Shop which asks for further information such as “how much do you want to pay” or “which store do you want to collect from”. The search is then narrowed down until the shopper confirms that they want to reserve an item. A notification is sent to their phone. By opening the notification, the shopper can confirm their order.
Argos, Chief Executive John Rogers said at the time of the launch in September 2018 that voice technology has the potential to “revolutionise how we shop in the future”.
He continued: “Argos is a digitally-led business at the forefront of technology and it’s really exciting that we are harnessing the simplicity of voice ordering with the convenience and popularity of click and collect to make our customers’ lives easier. We predict that the Voice Shop service will be a big hit and we will develop and refine the offer further as we get feedback from our customers.”
Voice Shop is a great example of how voice search can lead to complex queries which require not only for natural language to be understood but also for the search engine to understand the context for follow-on questions.
By working with Google Assistant it also highlights how voice is becoming increasingly important as an interface to wider applications when we are out and about. Most voice search queries are influenced by geography and time, according to Alari, meaning voice searches are often “Where is the closest XX?” or “What time does XX close/open. It doesn’t make sense then to restrict shoppers to only being able to talk to your brand via a fixed smart speaker.
It does mean though that retailers need to have content in a well-structured way so that the information is available in a format that a search engine can pick up in order to answer shoppers’ questions. “Retailers should be doing that anyway as it’s good SEO,” says Chris Atwell, CEO of Search Laboratory.
This is further complicated by the fact that searches are rarely done in isolation since an initial search will generate answers that provoke thoughts and often prompt further questions.
So, for example, if a shopper is looking to buy a set of speakers they may first search for the top 10 speakers and then follow this query by asking “which is the loudest”. This could then be followed by a further question around decibel output. Through a series of contextual questions, the querent is narrowing down their options to discover the right product. If shoppers are searching in this way using voice search they are typing searches in a similar way too.
These questions may be across devices too so “retailers need to be ready with strategies that allow them to provide answers on screen or through the voice of the assistant,” comments Alari, pointing out how searches via a smart speaker tend to give one result, so it’s important to answer specific questions a customer may have in order to be ‘the’ answer.
Retailers should be eying up the knowledge box – or position zero – at the top of the results on Google already, explains Atwell. It becomes increasingly important for smart speakers as this is the response which will be spoken back to the querent so the information needs to be on the retailer’s website in a readable format. Speakable on Schema.org, for example, shows how this can be structured and how a paragraph of text should be marked up so that it can be read out by a device, explains Malte Landwehr, VP Product, Searchmetrics.
“The difference between appearing as a result in a search screen and the result a smart speaker provides is significant. To achieve the best results retailers must use data intelligently, taking in contextual information of who, when, where and why a question has been asked, ensuring that their brand is the right result at the right time.
“Right now I’d say it’s more important to understand where your brand may be most useful and get these basic capabilities right than to answer every need in every scenario,” he adds. “The key to optimising for voice search is knowing what the questions are likely to be and why your product is the answer.”
The use of natural language also means that more searches are unique – and this has been a big learning curve for chatbots and shopping assistants utilising artificial intelligence.
“For customer services and customer retention and the after-sales market, it’s interesting to have skills on all the important voice devices which in Europe right now are Google Actions and Alexa Skills,” says Landwehr.
But, when it comes to enabling a smart speaker to interact with a retailer via an Amazon Skill or Google Action, Landwehr explains that retailers will need to be wary of the voice which they record or choose to use since that will become associated with their brand.
He does point out though that while smart speakers are proving useful for customer services applications such as queries around “where is my parcel,” they are not so good for purchases. Even being able to see which toothpaste is being added to a basket would be helpful to the customer as talking them through three different options and waiting for the shopper to make a choice would be arduous. Hence the move to smart speakers which combine voice with a screen – and the move by retailers to work with Google Assistant on smartphones.
Fashion company Asos is another retailer which has enabled customers to shop via Google Assistant on a smart speaker or via their phone. Enki, the Asos bot, helps customers to discover products and shop across 6 categories of men’s and womenswear. While the interaction is via voice or text, products can also be seen on the smartphone’s screen, thus breaking down one of the disadvantages levelled against smart speakers, the fact that the search results are not easy to browse.
As Alari concludes “Voice search is moving consumers from inspiration to action faster than ever before. For brands and retailers, the future is the removal of any friction along the way. Through being responsive and genuinely helpful, your marketing can be all ‘pull’ instead of ‘push’ – because you’re in the right place with the right offer, to the right person at the right time. In other words, you’ll be truly assistive – which is what today’s customers want, expect and value.”
However, should retailers be thinking about the next steps combing voice search with visual not just via the search engines but also on their own sites? The market for voice commerce will be large – reaching over $80bn a year by 2023, according to Juniper Research – but this includes money transfer and the purchase of digital goods. “We expect the majority of voice commerce to be digital purchases, until digital assistants offer truly seamless cross-platform experiences” says research author James Moar. “Connected TVs and smart displays are vital here, as they can provide a visual context that is lacking in smart speakers.”
Marks & Spencer launched Style Finder on its mobile site in January allowing shoppers to upload a picture or take a new one and be shown similar outfits that are available across M&S men’s and womenswear. The Style Finder tool, which has been developed with visual search company Syte, uses artificial intelligence to display results which are the closest match by style or colour to the original picture. Customers can add additional filters to help them find the perfect product based on personal preferences, such as size, price and colour.
“We know our customers are busier than ever and are often most inspired when they’re out and about. Style Finder helps customers instantly find what they’re looking for, without the need to manually search and filter through our products. Enhancing the customer experience is central to our digital transformation journey. This is a brilliant example of how we’re becoming more relevant, more often, to our customers who are increasingly shopping online and in particular using mobile devices,” says Jim Cruickshank, Head of Digital Product and UX at M&S.
Some 75% of all M&S’s online visits are from mobile and tablet devices.
Farfetch is at it too, launching a visual search tool at the end of last year. And it’s not just retailers. The search engines and social media platforms – Pinterest, Google Lens and Snapchat – have been turning to visual search too.
Snapchat’s new visual search functionality allows users to Snap an item or barcode and then be given shopping options on Amazon.
Google’s Style Match works with Lens allowing users to search via their camera. By pointing the camera of their smartphone at a certain object, users are provided with a series of results of similar items. “If you used Style Match when looking at a dress, you will be presented with a list of similar dresses (or, the actual dress itself) which are available to purchase directly if available through Google Shopping. This feature is designed to provide a quick and easy way to find items that you can’t quite summarise into a search bar,” says Alari.
Landwehr believes that “Pinterest has a lot of potential to become the visual search and visual decision-making engine for the whole funnel” since it can move visitors from investigating styles that they like right the way through to viewing individual items, such as lamps and clicking through to make a purchase. Pinterest has 250m visitors per month and everything on the site is based on images, inspiring visitors with everything from recipes to DIY and home design.
Its Co-founder Ben Silbermann told CNBC Closing Bell in April 2017 that computer vision technology is going to be “a big deal”. The future of search “will be about pictures rather than keywords,” he said.
It’s something that Gen Z say they want and the majority say that how a retailer’s website looks impacts their purchasing decisions.
However, everything again comes back to context; what is the person’s intent with taking this picture? Do they want to see a similar dress to the one shown in the picture? If it’s a picture of a room, are they looking for colour, furniture or information on the designer or who used to live in the house? Rather than bringing up an image of an item of food, should an app be returning recipe ideas, information on how and where the item is grown – if it’s an apple for example, or prices for buying said apples in the different supermarkets?
For the shopper it should be easier than typing into a search bar, scrolling, clicking and pinching of fingers to find the right information or perfect product.
There’s no questioning that voice and visual search are growing, albeit the latter at a slower rate. Shoppers want convenience and if a search engine can understand me when I say I want a black dress like the one in the picture but in my size and with long sleeves I will be happy.
This is still the premise of the human personal shopper since they can understand multiple inputs, context, emotion and can use digital tools to see stock levels. The user experience is everything and customer expectations are continually changing; getting it right for complex searches is going to take a lot more than saying OK Google.
Alessandra Alari, Head of Search and Digital User Experiences, Google UK, explains how retailers can best adapt their sites to be discoverable on voice and visual searches.