With voice searches on the rise, how can you make sure your content hits the right pitch? We asked iCrossing UK’s head of search, Mark Williams and head of strategy, Allyson Griffiths to share advice on how to research, produce and report on content for voice. But first, why all the noise?
Since the early speech recognition days of Audrey – a ground-breaking 1950s’ system that was only able to understand digits – technology has come a long way. According to Xuedong Huang, technical fellow at Microsoft, today its speech recognition technology is as accurate as human transcribers, with an error rate of just 5.1%. And as voice search systems become more sophisticated, their popularity is likely to keep growing. Mary Meeker’s 2016 internet trends report showed that “queries associated with voice-related commands have risen more than 35x since 2008”, and ComScore predicts that 50% of searches will be voice searches by 2020.
Research by Think with Google found five reasons voice assistants have shaped consumer behaviour:
They make life easier, allowing users to multitask.
They’re always available, as many people keep their devices at the centre of the home.
They’re part of the daily routine – whether it’s for checking train times or making a shopping list, 72% of owners said they use their device daily.
They’re so ‘human’, it’s like talking to a friend.
They’re a source of helpful information. Voice search device owners said they’re open to receiving the following information from brands: deals (52%), tips and life-hacks (48%), event information (42%), business information like store location and opening hours (39%), and customer service (38%).
Recent device hardware updates have also contributed to the uptake in use, with both Google and Amazon expanding their voice search device offerings at the end of 2017. Google added Mini and Max to its Google Home offering – allowing it to meet different consumer needs for product size and speaker quality – and Amazon now has Echo Dot, Amazon Echo, Echo Plus, Echo Spot and Echo Show.
"The interesting development from a hardware perspective is the addition of screens on the Echo Spot and Show, which opens up new possibilities for how we use these devices,” Mark Williams, head of search, says. “Both Google and Amazon are also working with third party hardware manufacturers to integrate Alexa and Google Assistant into other products (like Sonos and LG’s smart appliances)."
From a software perspective, Google and Amazon are continually learning from the data that’s being collected, but are relying on users to add Alexa Skills and Google Actions (the voice search equivalent of apps) to their libraries. This allows users to improve outputs for more niche search results, like checking their Fitbit with Alexa.
And it’s not just Google and Amazon making waves in the voice search world. Spotify confirmed that it’s testing a “new voice search interface” that lets users ask the app for the music they want to hear, making handsfree listening possible.
Head of strategy Allyson Griffiths explains that currently, voice search is mainly used for non-commerce related searches:
"These include navigational queries, like playing a song, or informational, like a quick answer to a question. From a customer purchase journey perspective, the use of voice search is still maturing. Amazon has led the way here, with consumers using Alexa to re-order favourite products where there’s little need for comparison or reading reviews."
According to the 2018 Amazon Shopper Survey, 36% of 55-64 year olds “prefer to use Amazon Echo or Dot to search and shop” rather than a mobile, desktop, laptop or tablet. Uptake among younger shoppers is strong, too – almost a third of 25-34 year olds surveyed also said they’d rather use these Amazon voice search devices.
Aside from the usual SEO best practices that make content readable for search engines, there are several ways we can up the chances of content ranking for voice search.
Mark suggests thinking about voice search with a question-based approach; finding long-tail keyword phrases that include question words, then building content around these queries. Start with…
Answer The Public, a question-based keyword research tool that suggests related queries.
Google AdWords keyword tool, including ‘who’, ‘what’, ‘where’, ‘when’, ‘why’ or ‘how’ along with the product or service.
‘People also ask’ boxes on Google to research additional common questions.
Social listening tools – particularly useful for more up-to-date trends and language and tone insights.
FAQs – mine your customer service teams for common questions to understand what consumers care about.
As well as keyword set research, consider:
Audience tone – as voice searches are more conversational, they tend to use more words than traditional keyword text searches. Google reported that 70% of requests to its assistant use “natural language”, which shows that many searchers are speaking to their voice assistant devices the same way they would to a friend.
Competition – voice searches (especially assistants) use only one result to answer questions, so if competition levels are already high, it may be worth considering other topics to pursue.
When it comes to writing content for voice search…
Be comprehensive. Rather than thinking about the length of the answer, make sure it’s as comprehensive as possible by grouping keywords and topics. Google and Bing (which powers Alexa) compare every answer to the question, looking for the most authoritative and complete.
Speak the searcher’s language. Keep your content natural, conversational and avoid jargon. Make sure your copy contains the long-tail query phrases found at the research stage, as well as alternative ways of wording these phrases to capture different search variations.
Use structured data. While search engines can read data, they can’t always understand exactly what it means. Structured data is organised through a mark-up, which adds additional information about a webpage in a way that search engines will understand, to spell out what the content is about. Google is more likely to use copy in its featured snippet (which is often used in voice search responses) if it’s clear on what the content means. In terms of new developments, there are a few schema types like ‘speakable’ and ‘SpeakableSpecification’ pending on schema.org, which could potentially be used to support voice search.
Consider the clearest layout. Search engines like lists, numbers, steps and tables – any structure that’s easy to segment and pull out for a featured snippet – and good content discovery should identify trends in effective structure by sector and topic.
So after carefully researching, planning and crafting content for voice search, how can we measure its success?
As Mark explains, many packages struggle to filter voice searches from standard text searches: “Currently we need to filter inbound keywords for ‘who’, ‘what’, ‘where’, ‘when’, ‘why’ or ‘how’ and then assume a percentage of these might be voice searches, to show the traffic/results from this medium. Although it’s worth noting that with most voice assistant results, users don’t actually visit your physical website, which limits our ability to track results.”
While you can see if someone’s initial query (before they clicked through to your site) was made via Google Assistant in AdWords’ search query reports, Allyson adds that we can more accurately measure results “via uplifts in other areas, as we do with any marketing channel where we can’t directly track activity impact.”
The uptake in voice search queries is good news for…
Businesses with retail stores – Search Engine Watch revealed that “mobile voice-related searches are three times more likely to be local-based than text.”
Those selling fast moving goods, like groceries, with high repeat purchase rates
Anyone selling through Amazon
Not so great if you’re a business that currently generates revenue from advertising, as Allyson explains: “Increasing voice search volume and decreasing keyword searches will lead to fewer ad impressions, unless a viable ad solution for voice is introduced.”
Over the next five years, Mark predicts major improvements in the way search engines understand and respond to voice searches, with:
Fewer questions that voice search engines can’t answer.
More personalised results.
No need for users to include context in their questions, or for queries to be asked in a specific way.
Search engines able to chain questions together to improve context.
But will the rise in voice queries have a direct impact on text-based desktop and mobile searches? With the uses for voice so specific, it’s unlikely we’ll see any dramatic decline in the immediate future:
"I personally expect voice search to continue much as it is today – useful for hands-free, short, quick information, but not competitive with desktop/mobile experiences in terms of influence on a customer journey,” Allyson says. “Until there’s a good brain computer interface, of course, when we’ll be able to ask our assistant to display visual information in our eyes…"
Look forward to seeing you there…
We believe that moving too slowly in digital is the biggest risk your business faces. If you are ready to move faster in digital, we are here to help.