Voice search SEO: The Ultimate 7-Step Guide

Contents

– What is voice search SEO?
– What’s the difference between voice search SEO and voice command?
– What percentage of search is voice?
– Is voice search the future?
– Voice search SEO terms decline over the last three years
– Why do you need to conduct voice search SEO?
– *NEW* Brand new voice search findings
– So, why focus on voice search SEO now?
– How does voice search work?
– How do I optimise for voice search?
– – 1. Conduct the right keyword research
– – 2. Target rich SERP features
– – 3. Use structured data
– – 4. Make your content mobile friendly
– – 5. Integrate with assistants using Google Actions or Alexa Skills
– – 6. Check your page speed
– – 7. Optimise for local search
– How can you measure your performance in voice search?
– Conclusion

First of all, let’s cover the basics. If you already know your stuff when it comes to voice, feel free to skip…

What is Voice Search SEO?

Voice search SEO refers to the process of optimising content with the correct structure, markup and long-tail keyword phrases, so that it stands a better chance of being read aloud by a voice assistant.

What’s the difference between voice search SEO and voice command?

The sole difference between voice search and voice command is intent. With voice commands you generally want to perform an action – for example turn on a light, listen to the radio, or call a relative – whereas, with voice search you’re often researching, browsing, or shopping.

While voice command is important for you to consider, voice search presents you with a significant commercial opportunity, as you can meet your customers at every point in their journey, and encourage them to convert. And that’s why, in this guide, I’m going to be focussing solely on voice search!

What percentage of search is voice?

The stat ‘50% of all searches will be conducted via voice by 2020’ has long been the mantra of the SEO community (people say this was Comscore – but it wasn’t – it was Andrew Ng [above] and he was referring to searches in China). That means voice searches will be surplus to the 3.5bn searches we see today – to the tune of 50%. This will undeniably offer a whole new realm of marketing opportunity.

Is voice search the future?

Voice search is fast becoming a staple part of our everyday lives, and could well be the future of search. Worldwide, more than 420 million voice assistants have been sold, and 2.71 billion people own a smartphone in 2019 – the majority of which have voice capabilities. In fact, 20% of the searches on a mobile device are voice-based (Global Web Index), and as voice develops, that figure will only grow.

Voice search SEO terms decline over the last three years

Interest in ‘Voice search SEO’ (a category made up of terms that marketers and content creators would themselves search to discover how to optimise for voice) has declined over the past 3 years; falling from a peak of 6.8k searches in January 2017, to 1.8k searches in August 2019. This could mean one of two things; that we have collectively lost interest in voice search, or that we believe we are sufficiently optimised for it.

Either way, this is far from ideal, because while interest in optimising for this topic is declining, voice search developments are very much powering on.

And it’s not just Google – Baidu is on top of this challenge as well, and has focused heavily on improving its speech recognition accuracy. In 2017, when Google celebrated reaching the coveted 95% accuracy threshold for voice recognition (putting it on a par with human speech), Baidu was at 97% – and is now gunning for 99%.

If you thought you were prepared in 2018, chances are (sorry to be the bearer of bad news) you’re probably not anymore! As with SEO, voice search optimisation is very much an ongoing process. We can’t afford to get complacent!

Why do you need to conduct voice search SEO?

Traditional search, as it stands, underpins the entire customer journey, and now so too does voice. As such, it affects all industries and presents Search Marketers with an unmatched opportunity to meet their customer at valuable touchpoints.

In our new research we have made a critical discovery.

The results for voice queries are very different to equivalent searches using plain old keywords.

What that means is you can’t just rely on your traditional marketing optimisation to reach your customers when the voice search switchover occurs.

Alan Ng (Adjunct Professor of Computer Science at Stanford University, formerly at Google and Baidu) predicted on Twitter that as soon as voice recognition technology reaches the 99% accuracy threshold, “We’ll go from barely using it to using it all the time”.

Brand new voice search findings

And, in fact, our new research suggests that this may already be happening, with people stating that they are 26.5%more satisfied with voice searches, and have a higher intent to purchase via voice than traditional search.

So, why focus on voice search SEO now?

Well, aside from the popular 2020 predictions – there’s been a Google Algorithm Update recently that I’m convinced will see the rebirth of voice search.

Voice has been routinely propelled forward by algorithm updates since the initiation of Google’s Hummingbird in 2013, which redefined Google as a sophisticated ‘Intent’ and context matching engine. This was ramped up in 2015 when RankBrain was introduced, and ever since, both have been instrumental in the drastic optimisation of SERP relevancy, which has subsequently enabled the voice assistants we use on a day-to-day basis to become increasingly accurate and intuitive when answering our questions.

But now, with the arrival of the BERT update – an algorithmic development which makes it easier for Google to interpret full sentences (i.e. its biggest search improvement in five years) it’s looking like the voice search search tipping point is about to topple, and we may see Google switching to prioritising voice results in the very near future!

How does voice search work?

For a voice search to ‘work’ – i.e. understand any query and provide a relevant, coherent and user friendly response – it requires Natural Language Processing, which is baked into Google’s algorithms (Hummingbird & Rankbrain). NLP analyses 5 key components of a single voice query to be able to serve a relevant and coherent response. These are:

• Phonology – speech analysis
• Morphology – word choice & word formations
• Syntax – sentence structure & grammar
• Semantics – word meaning & context
• Reasoning – Planning

And with this kind of processing, Google is able to do amazing things. Here’s one of my favourite examples of Google being able to understand pragmatic language – i.e. language that can only be understood in context (think of the word ‘Thing’. You can never understand what that word means, unless you interpret the connected meanings of the words surrounding it)

And finding meaning is not the only thing Google’s algorithms can do. They can understand your accent, your dialect and – as we’ve seen above – even remember some of your most recent questions, to give you an even more refined response to your searches.

How do I optimise for voice search?

1. Conduct the right keyword research

TL;DR

1. Use conversational language – Move away from keyword stuffing

2. Theme content around long-tail and niche queries, over 5 words – Quora does a fantastic job of this.

3. Create content in a Q&A format

4. Answer your target question within the first paragraph of your content – And optimise other traditional on-page SEO elements with your voice search phrase

5. Incorporate plenty of keyword and phrase variants in your content – to have a greater chance of being read aloud for synonymous questions

6. Explore People Also Ask (AlsoAsked.com) to identify questions

7. Explore Google Auto Suggest (AnswerThePublic.com)

8. Create FAQ pages around the questions most important to your business – Benu Aggarwal suggests creating FAQ for every major trophy page on your site.

9. Involve other teams in your research – Including Customer and Support, to get a deeper insight into the questions users are asking to find you

10. Measure search volume of relevant questions – inject these questions directly into your content.

A big difference between traditional search and voice search is the importance placed on keywords.

The natural language used in voice search shifts the focus from keywords to semantics – moving from statement based searches to full sentence questions.

“70% of requests to Google’s assistant are expressed in natural language rather than keywords”

*Google*

You’ll need to think in terms of conversational language instead of typed.

You’ve got to perform for long-tail keywords (download our brand new research below, on how different voice queries are to traditional queries) to have a chance at voice search visibility. This requires different keyword research to traditional SEO, as you need to understand what the full intent of the user is.

In traditional SEO you may assume that someone searching for a ‘black dress’ is looking to buy a black dress. Whereas in voice search SEO, it may be the user asks ‘where can I buy a cheap black dress’ which offers far more insight into their shopping intent.

These are a couple of my favourite resources that help to identify the sort of questions people are searching.

– AlsoAsked.com scrapes the questions that return in Google’s People Also Ask SERP feature for the particular topic you’re optimising for.

– AnswerThePublic.com gathers all the Google Auto Suggest questions that appear in the search bar before you even complete your query.

Taking these scraped questions and measuring their search volume will narrow down which long-tail keywords you should look to perform for. Then use your chosen questions as headings, and accompany with clear and concise explanations.

2. Target rich SERP features

TL;DR

1. Answer ‘How’, ‘What’ and ‘Best’ queries – These are most likely to trigger voice search snippets such as the ‘Answer card’

2. Google uses SERP Feature content for voice search responses

3. Track SERP feature visibility – discover the feature landscape for your target phrase – are there any voice features you can optimise for?

4. Mimic the format of the existing snippet appearing for your target query

5. Study the competitor skyscraper snippet that’s already appearing

6. Read my full guide on how to own a featured snippet here

Building on what I’ve said in step 1, targeting SERP features is your best bet for successfully performing in voice search.

SERP features are the larger, usually more visual results returning in modern Google. These take up more space in the SERPs and display far more information.

When it comes to voice search, these features are super important – particularly the featured snippet that sits at the top of the SERPS, also known as ‘position zero’. Appearing even before paid adverts, the featured snippet can include copy, bullet points, images and even charts.

A voice assistant will pull the answer to a spoken query directly from this snippet, so having visibility here should be the top of your list when it comes to voice search performance.

One thing to bear in mind when attempting to return in voice search, is the kinds of queries that are most likely to return a SERP feature. According to SEO Clarity, more than 20% of rich snippets are triggered by these keywords…

All of these words are either used as part of a question, or imply users are searching for a list. This is invaluable to know!

So, how do you get into SERP Features for voice search?

Once you’ve decided on your focus ‘How’, ‘What’ and ‘Best’ queries, how do you go about landing that SERP feature?

Track SERP feature visibility

To rank for a SERP feature you firstly need to assess the opportunity available across the landscape for your focus voice searches. In the example below, we’ve analysed the SERP Feature landscape for voice searches around ‘Getting a six pack’ for November, 2019.

What we are looking at here is the rich ‘SERP Radar’ within the Pi’s SEO Platform, which shows a % of feature presence in the SERPs for a group of voice search queries.

In this example, ‘Video cards’, ‘Answer cards’, ‘Classic links’ and ‘People also ask’ features return for these voice searches.

You are also able to see the % of site presence. In this example, Coachmag.co.uk is appearing for 100% of ‘Classic links’, but for the feature that is most likely to be read aloud in a voice search (‘Answer cards’), it only has a 0.9% presence out of a possible 86.6% – so it can by all means do better!

*

We can also track feature visibility for a single voice search query to see how the SERP Features change, and which features take highest priority in Google.

In this case, we’ve focussed on ‘How to get a six pack’ which, despite being long-tail, manages to accrue a healthy 20 searches pcm.

The ‘SERP Matrix’ (above) shows us the SERP layout of Google (each column) over time (in this case – every day).

We can see above that ‘Answer cards’ (in black) are indeed the highest ranking feature on Google, meaning there is definitely voice search opportunity here for our query ‘How to get a six pack’!

Check you are ranking on page one

Before you go any further, check you are ranking on page one. You have a far higher likelihood of appearing in a SERP feature if you’re already ranking on the top page.

If you aren’t, you need to address this issue first before going any further.

In the example above, Coachmag.co.uk appears in position six on page one, so it can progress to the next step.

Mimic the format of the existing snippet appearing for your target query

The next step you need to take is to discover the competitor already appearing for your focus voice search.

Healthline.com owns the ‘Answer card’ here, as we can see below.

When we click through into the SERP (see below), we can see that our competitor’s content takes the form of a list, so we know our content would have the highest likelihood of stealing position zero if we include bullet pointed content.

Study the competitor snippet that’s already appearing

We’ve proven in a study that you can usurp your competitors in the Google Answer Box, so what can you do better than them to take that coveted voice search position?

Improve your schema? Provide the answer more immediately, or simplistically?

Study their snippet and start optimising.

3. Use structured data

TL;DR

1. Create a sitemap in Google Search Console – to ensure your site is easy to crawl

2. Schema markup may aid your performance in SERP features – especially in PAA

3. Indicate to Google where your voice search answers are with schema

As we’ve explored, Google SERP features or featured snippets populate ‘Position zero’ in the SERPs – a prime location that provides the content that gets read aloud in a voice search.

As you probably also already know, only one answer can be read aloud by a voice assistant, so optimising your content for these features is absolutely critical. So how you do that? With structured data!

Voice users frequently search in a question/answer format. If the search engine has a clear understanding of where the answers lie in your content, as directed by your metadata, it is more likely to accurately showcase your content.

Structured data quite simply tells Google what your content is, without affecting how users see it. You can markup the following information, so that Google ‘gets’ what your content is about:

– Articles
– Q&As
– Restaurants
– Movies
– TV
– Software
– Events
– Local businesses
– Reviews
– Products
– Recipes
– People
– How to content

Once you’ve marked up your chosen content you can then test it with Google’s Structured Data Testing tool.

It’s not explicitly known if schema markups aid in performance of SERP features, however there have been a fair few mark-ups introduced recently that imply they do – question markups in particular could potentially help performing for the People Also Ask feature.

Interestingly there is a new schema, in its beta stage at the moment, called ‘Speakable’ which is specifically for any copy that you want Google Assistant to speak. However, this is only available for ‘valid’ news sites at present, identifying the schema in the ‘Top stories’ feature – but this may evolve in years to come. Definitely one to watch!

4. Make your content mobile friendly

TL;DR

1. 52% of smartphone owners use voice assistants

2. Optimising for mobile gives you the opportunity to feature in voice search

3. Improve your mobile page speed with AMP – having this enabled means you’re more likely to perform for SERP features, which is extremely important for voice search visibility.

Whilst you may associate voice search with smart home devices, most voice searches actually come from mobile. People carry their voice assistants with them wherever they go!

As the majority of web traffic now comes from mobile devices (57%), it’s more important than ever to pay attention to your mobile content design, even if voice search isn’t a priority.

However, there’s some handy overlap between optimising for mobile and voice search optimisation.

For example, improving your mobile page speed. According to Backlinko, voice search results load 52% faster than average pages.

If you use the open-source initiative, Accelerated Mobile Pages (AMP), to increase the speed and readability of your mobile content, Google is more likely to serve your site.

As such, AMP increases your potential to perform in SERP features on mobile and, as we’ve already discovered, the majority of voice queries are answered with content from SERP features.

But if you’re still not sold on using AMP, this function will also help you create Google Actions (see next section) – it’s a no brainer!

5. Integrate with assistants using Google Actions or Alexa Skills

TL;DR

1. Actions and Skills enable users to have conversations with your app / content

2. Both enable you to submit functions to be enacted by a voice assistant

3. Functions can be anything from turning on a light to playing a game

4. Schema markup in Google search can auto-generate Skills on Google Assistant

5. You can submit Google Actions for your content without coding knowledge

Remember when we defined the distinction between ‘Voice search’ and ‘Voice command’ earlier on?

Well Google Actions and Alexa Skills are more concerned with the latter of the two.

However, applying schema markup on Google SERPs can enable you to perform better for both voice searches and voice commands, as marked up content also has the ability to auto-generate actions for the Google Assistant.

Google has this to say on the subject…

“In addition to presenting your content as a rich result on Google Search, we can read your markup to automatically build an Action that presents your content on the Google Assistant.”

Using these developer platforms, you can boost awareness by generating fun or useful functions related to your brand, which get performed by a voice assistant.

For example, dating site Eharmony has built an Alexa Skill which enables you to ask Alexa about your matches, while content creators on Google can markup their how-to video content, to get a step by step version read aloud to searchers via the Google Assistant.

If you want to create content-based Google Actions, and you already have how to videos, recipes, podcasts, or a news article marked up with structured data / Accelerated Mobile Pages (AMP) then you simply need to claim your Actions through the directory.

But if you do choose to create a skill or action, make sure it is useful and aligned with your business goals!

6. Check your page speed

TL;DR

1. Users will not wait for a slow site

2. Check your site speed with PageSpeed Insights

3. Mobile site speed is more important that desktop for voice search

The importance of page speed to voice search is simple. Often consumers are on the go and in a rush, and they don’t want to be waiting around for a page to load.

I touched on mobile site speed in point 4, which is actually far more important than desktop speed when it comes to voice search, so prioritise that.

You can test your site speed using Google’s PageSpeed Insights – the tool analyses your current speed and even suggests ways to improve it.

7. Optimise for local search

TL;DR

1. Claim your Google My Business listing

2. Feature your address on every page of your site – this will give you a greater chance of appearing in ‘Near me’ voice searches

3. Add schema markup to your site

4. Provide opening times – To answer local research questions

5. Encourage users to leave reviews – To provide more information about your brand that can be surfaced in voice

6. Do geo-local keyword research – i.e. What are the best restaurants in Brighton? Be sure to adopt a ‘Voice query research’ mindset and answer local questions

Local search makes up a substantial 22% of voice searches. This makes sense, when you consider that people constantly use their mobile phones on the go.

We reach out to our voice enabled devices for assistance without a second thought – a practice which Google has famously coined ‘Micro-moments’ – and a huge proportion of these moments are local.

In fact, according to a Search Engine Watch guide on voice search and local SEO, there were 1 billion+ voice searches a month in January 2018, and 40% of those mobile searches had local intent.

“58% of consumers have used voice search to find local business information in the last 12 months and 46% of adults use voice search for local business searches daily.”

*Bright Local*

A relatively easy first step for you to take would be to claim your ‘Google My Business’, google’s answer to the Yellow Pages that are currently buried deep in your attic.

Through taking ownership and verifying your company with GMB you are able to take control of your Maps and Search, allowing you to provide precise information with tailored keywords that will stand you in very good stead within local voice search.

A common local and generic voice search query is “things to do near me”, which has increased 6 fold over the last two years.

By providing opening times and extensive reviews to Google, your site is more likely to appear in position zero for this narrow scope search criteria.

Something which is very important to note is that searches with local intent do not show up in a Google Answer Box. According to Sherry Bonelli at BrightLocal “Even if Google isn’t including local-related voice searches in Answer Boxes right now, you should prepare for it when they do.”

With the advancements of BERT’s natural language processing, will this be the next-step for Google?

For more information on local voice search usage and demographics, check out BrightLocal’s brilliant local business study.

How can you measure your performance in voice search?

The easiest way to do this is on a category or topic level. Assuming that you already have your search term universe established the next bit is easy. Begin with a sample of keywords and for each one, come up with one or more voice equivalents.

For example, ‘Men’s designer skinny jeans’ for the keyword term might have a voice equivalent – ‘Where’s the best place to buy a pair of designer mens skinny jeans?’

In the most recent piece of research that we carried out, we showed people images of a selection of items of clothing as well as accessories. We then asked them to perform regular queries, followed by voice alternatives; all of which we recorded. However, with a bit of common sense, you should be able to come up with the voice equivalents without a focus group!

Once you’ve done this, all you need to do is analyse your organic performance for the keyword group and compare this with the voice group. In Pi you can do this by creating a Comparative Visibility Index. This quickly gives you a feel for any gap in the performance of your content for regular keyword queries.

It’s important not just to take a snapshot but to monitor any gap over time to get a feeling for whether your content is getting more or less ‘voice friendly’

Here’s an example Index:

Based on around 250 queries related to high street fashion items, we can see that ASOS performs very differently for standard keyword queries:

So, what does this mean for ASOS?

Well, if we all woke up tomorrow and started using voice, then ASOS’ organic visibility would drop through the floor.

The impact would be catastrophic, probably worse than the drops they saw last year. Of course, that’s not going to happen over night, but, if day-by-day and week-by-week we start to adopt voice and we start to use longer, slightly more ambiguous search criteria, confident that Google’s RankBrain and BERT are effectively disambiguating and returning more relevant search results that better match our intent, then the same inevitable result is likely.

As the old adage goes – fail to prepare and prepare to fail!

Conversely, when we look at Amazon for the same searches (i.e. voice and keyword) their visibility gap is almost non-existent:

Well, it would seem that these guys are doing something right! Is this intentional? I doubt it! I turned the clock back 12 months and apart from a 30 day period in April/May 2019, their performance for voice has matched their performance for keywords.

Conclusion

With rapid adoption, growing global comprehension and the significant advancements of Natural Language Processing algorithms like BERT – not to mention the findings of our brand new research – there is no doubt in my mind that voice search is going to become a huge priority for us Search Marketers.

If we become complacent, as it seems we are at risk of doing (looking at the decline in interest around ‘Voice search SEO’ in the SERPs), then we could be in big trouble when the SERPs start to prioritise voice.

It is vital we get prepared now, by optimising, monitoring our visibility and staying abreast of new developments – or else we could pay for it in the very near future!

Thanks for reading! For more tips and takeaways like this, you can do one of a few things:

Subscribe to Pi Datametrics’ YouTube channel for my upcoming videos – which I’ll be posting regularly!
And, finally, follow me on Linkedin or Twitter