How to Optimize your Website content for Voice Search – #CrawlingMondays 18th Episode
Learn how you can become the answer in the Google assistant for voice search queries with your existing Website content as well as to expand your presence by creating Google Actions.
Today’s edition of Crawling Mondays is going to be about a very hot topic. Voice search. This is a topic that I have spoken about at a few conferences already, and then I realized that I didn’t have a Crawling Mondays episode about it, which I’m going to fix right now because it’s a very important topic to clarify a few things.
And also to give you a few resources, to start optimizing as much as possible for it. So, yes, voice search, or conversational search, better, is a little bit everywhere, because it cannot only be found through the devices that got very popular since a couple of years ago.
But, realistically, it’s now everywhere. The web search, the app, there has been a little bit of a buzz about it because there was some predictions saying that half of the searches were going to be by voice by 2020. This was taken out of context. It was based on some data, some predictions made in 2014.
And it wasn’t only going to be about voice, or speech search, but also about images search. Also, there’s a few people who are scared about the effects on CEO, and there are a few things to clarify about that. I actually did this little poll a few weeks ago, before my presentation about this topic at SMS London a couple of weeks ago.
And I asked my followers, very techy-oriented people, “Do you search in Google via voice at the moment?” The answers that I got was like 58 percent of the people from 1,100, a little bit more than 1,100 people who answered, they never search through voice.
33 percent, only once in a while, and nine percent, yes, all the time, and then I got a few clarifications about it, saying, like Izzi Smith, “My Google Home device is only used for setting cooking timers and song requests.”
And then Shure said, “I’ll use it for the map apps like twice a year.” And then James Crawford answered, “The only time that I use it now is for entertainment.” And, realistically, even Google share a little bit of data about it. Like, also a year ago or so, saying that voice is about action, and most of the requests that they are getting at the moment are on one hand very action-oriented, for simple type of requests.
They are not necessarily related to actual web search, and then on the other hand, screens change everything, and again, theses are I’ve insights about voice technology that Google shared a few months ago, and they mentioned that nearly half of the interactions with the assistant today include voice and touch input.
Most of our queries are complex. For example, about reserving a hotel in London, or about buying new clothes. You are not very likely going to be satisfied by a voice answer, so yes, I can understand how there is a trend requesting, in a very natural way, through voice, many times.
Because we are on the go a lot, and we have the access to all of these devices. But realistically, in order to be satisfied from these answers, we need the visual output, and that is why, also, Google, for example, has released more smart displays, and I think it’s much reasonable to speak about conversational search era that we’re entering.
Where many of the input, the requests, the searches, can be done via voice, but for the output, a high share of this need to be visual in order to satisfy the customer need, and the connect … because at the end of the day it’s about this, right? Connect our content to satisfy their need, whatever the medium it is.
Let’s take a look how. How can you start organizing your content and focusing on these conversational type of queries? These are natural questions, now. The six Ws that are being used by your users to research for content. The what, when, who, where, why, how, and you can start gathering this data from your own content, actually.
You can go to your Google Search Console and use any of these terms, for example how, here, to identify how many queries are you getting right now, with how, and I can see here in this case, for remoter search, for my own website, “How to get remote jobs in India, how to make money as a digital nomad, how to use Frivolite abroad, how to become a digital nomad,” right?
And you can see the performance that these queries are providing you at the moment. The clicks, impressions, click to rank position. Of course, you can take a look at what is the content, where are the pages, the ranking for these terms, for these conversational type of queries.
If it is relevant, if it is meaningful enough. Of course, you can do this by using other tools, QR tools, like SEMrush, Ahrefs, Seestreaks, not only with your own website, but the one of your competitors, so you can go and take a look at how your competitors are already leveraging conversational queries.
They are ranking for the same that you are, or any other queries. So, for example, here, flex jobs, and I am looking here for the how queries. I can see, “How many hours is part time, how to fill out a resume, how to be successful without college, how to fill a block on a job application.”
So many queries, conversational queries, for which they’re already ranking. I can see their position, I can see an estimate of the traffic that these are bringing for them, I can see the pages. They are ranking what … I can see the search volume. That is not trivial at all, so this gives me a lot of ideas already of the type of conversational queries that my audience is using to search for my content.
And of course, I can use this as an input for a bigger keyword research. So, as a next step to identify even further opportunities, to target this conversational type of queries, you can use also your relevant topics as an input, and go to the questions filter in any of the keywords [inaudible 00:06:07] nowadays.
Pretty much all of them, whether in Ahrefs, or in SEMrush, Keyword Magic Tools, and Keyword Finder, for most often keyword tools nowadays, they will have the questions filter. So, for example, for remote jobs, I can see here only the questions regarding these topics.
And I can see here even more ideas and opportunities of conversational queries that I can target with my content of how to find a remote job, how to get a remote job. Like this, I will be able to see not only which hard terms the conversational type of queries are providing me and my competitors are at good rankings, but also even for your queries that I can target with my content and structure my content in a way that. we are going to see, we can optimize in order to maximize the chances to rank well as the answer through the Google assistant.
So, how do we become the answer for these conversational queries that we have identified and find relevant to rank for? It’s about becoming, also, the first result, because we are going to see that there’s a high correlation between them.
We want our more factual type of questions will be satisfied and answered by Google knowledge graph through the answer boxes. Here I ask through the Google app, the assistant, “What’s the name of Prince Harry’s son?” And I got the name, which is exactly the way that it works in desktop search, as we can see here.
So, for these factual type of queries that are satisfied by the Google knowledge graph, and for which there is an answer box provided, just going to be pretty much it. The answer through the Google assistant.
On the other hand, they’re also these queries, like for example localized queries, for which Google will provide the maps result, their own services features, that we are going to obtain as the answer through the assistant. We get it here in the desktop search results and we get it, also, here, as the answer. On the other hand, there are these much more complex type of queries. More transactional driven, customer-oriented driven, that are not satisfied by the answer boxes, but for which featured snippets are provided in many occasions.
We are going to likely want to target these ones, because these are the ones that we can uses to connect with our customer, within the customer journey. For these other more complex type of questions, for which a featured snippet is provided, we can optimize our content to become the featured snippet, and there’s going to be a high correlation.
As we can see that this is the one selected as the answer by the Google assistant. So, for example, and this case, I asked, “Where are the best neighborhoods to live in London?” And a featured snippet is shown here, and the same is provided here as the answer by the assistant. Right?
And there has been many researches, and this is taken from one done by SEMrush. 40.7 percent of all voice search answers came from a featured snippet, and we can see that many of theses are questions, prepositions, comparisons, and mostly these are targeted by FAQs, guides, resources, addressing informational type of queries in the customer journey.
So, for example, the best flat prices in London. This generates a featured snippet like this, and the content is taken from the different areas of this blog post that has been published in this real estate website. We can also not only target, though, informational type of queries, but also transactional ones, more and more, so for example, best prices for tires in the UK.
Or buy cheap laptops, for students, too. So, how we can optimize our content for that; first, already be ranking well for them, because the featured snippets are generated by the first results already. There’s a high share of them that are already ranking the first three positions, so that is the first step that we need to do.
Then the content should be organized in a very well structured way, in a very concise way, so Google can easily highlight the most prominent areas that actually answer the query, the conversational query from our users, and that these are areas that are going to be spoken to. We can see many examples here of how this content is well structured, it’s mobile friendly, of course, already secure, it’s already optimized, and it’s also written in a way that can be easily spoken.
And this is the other thing. A year and a half ago or so, Google published these best practices for evaluation of speech for the Google assistant. These are actually the criteria that Google takes into consideration to evaluate the content for voice, so besides the typical information satisfaction type of criteria that Google take into consideration for search results in general, and find information about the length, the formulation, the elocution, how this content is spoken.
Like, spoken answers must have a proper pronunciation across them. Even with examples, it’s not only the usefulness of the response, of the answer, but also the quality of the speech, the length, the formulation, the elocution, again.
And we can have, even, examples here, so I will say it’s definitely important for every content that we now write that we read it out loud, that we hear how it sounds, that if it is short enough, concise enough, if it is well formed, if this content can be easily spoken, on one hand, so definitely take a look at this resource here.
Then we can also use the how to structure data for rich results, and this will also generate how to actions for the Google assistant. This was something that was announced in the Google I/O a few weeks ago, so again, this is a win-win in these, all the cases, and we can see many examples of this in websites like Instructable, for example.
They have implemented the how to structured data for most of their pages, and they’re already generating the result for many of the queries that they target like this. If you’re a media site in the US, you can also use the speakable structured data.
Search Engine Roundtable’s already doing, by using this [inaudible 00:12:30] and then this is the thing, right? Even if you’re not a media site in the US, you can use the rules, the guidelines, that are provided by the speakable structured data, again, to format your content in a way that is very easy to be read, to be spoken.
So, content that may sound confusing in voice only, in voice forward, don’t add the speakable structured data to it. Learn that highlighting an entire article with speakable structured data, focus on keeping concise headlines and summaries that provide users with comprehensible and useful information.
Break up information into little sentences so that it reads more clearly for TTS. So, this is fantastic, again, because it gives you very specific guidelines to follow to structure your content, and you can use all of these criteria and structured data, also, to optimize your content to be concise, easier to read out loud, and well structured, as, for example again, there is another example of Instructable, for which they are generating this featured snippet.
It’s also provided as an answer by the Google assistant. And you can, of course, identify if you are triggering already this type of featured snippet result very easily, actually, by using again, SEMrush, Ahrefs, or even Seestreaks here. As I can see, any of these tools have a filter here for featured snippet, and I can see all the type of featured snippet that, in this case, Flight Shops is generating.
Like, in text, the videos here, and I can see which are this query. Something additional that you can do, of course, is to track this with your own run trackers. Most of run trackers nowadays also have the support to track sort of features in general, so for example, here, with 米6体育直播nbamonitor, for one of these terms that I am ranking with, I am shown that this is included as a featured snippet here because of the quote, and I can see even further not only the ranking but the actual picture rank, the visits that are generated from this result, and the behavior.
The conversion, the bounce rate, so I can also assess the impact we provide or we show as a featured snippet. To additionally expand your presence through the Google assistant, you can use Google actions. Google actions are conversational applications for which Google will find a matching 10 with the queries that their customers ask or search for.
And they will provide them as an answer, if there’s a match, so you can see, actually, a really good list here in the Google assistant directory listing a lot of actions from calculators, to quizzes, to news actions, etcetera.
A lot of websites use them as an additional support for users, or just to interact in a fun way with users, and to better provide answers to many of their queries, and realistically, until now, the most popular way to generate or develop these actions were through platforms like Dialogflow.
Right? That are used to develop a little bit of more complex type of conversational experiences, not only for the Google assistant, but also for Amazon Alexa. Even for bots. However, I understand that Google wants to facilitate the process, and what they have done is that they have informed that they will now generate, also, actions when we use structured data.
So, they will identify the usage of the FAQ structured data, the howto structured data, recipes, the news, and they will generate automatically actions when they find this as information that is marked up. However, we can, of course, make sure that we properly generate our own actions to have far more control with them.
And we don’t necessarily need to develop them by coding from scratch. There are easy ways to generate them, now, with templates. Actually, I created an action with a template last year for my website while my web traffic dropped, which is a checklist to identify the reasons for losing traffic, and I created this quiz at the end, and you can check this quiz actually in the actions directory over here, and the way that you can invoke the action here, the views of a few people, and this is the Google Docs, the template that I have needed to fill to generate the quiz.
Which is actually a flashcard, here, so here are the questions. These are the answers. This is the followup to generate the conversation, with the assistant with the users, and this is the configuration. I just have to fill this template and now, actually, Google is providing a way not only to generate quizzes or actions like this, but also to generate howtos which, I have to say, I can see that might be much more meaningful and useful to generate, for example, FAWs type of experiences, report type of experiences, to expand your current one.
And you can see in a very similar way, it’s just to fill these Google Docs, and then you will connect with the actions on Google console, so you can monitor your actions, you can upload your actions, do changes through your actions, et cetera, and you can specify how it will vote over here, how your action is built, the personality, the voice of your action.
You can test it with a simulator here, and you will get, also, data about the usage of your actions, which I think is super useful in order to better understand the type of interaction and behavior that your audience is having over the assistant. I will highly recommend you to check it out, to give it a go.
Generate games, flashcards, how-to guides, trivias or quizzes. I hope that this has been useful for you and that with the information that I have provided in today’s episode, now, you are not only a little bit more aware of what is happening with voice search and conversational search, but the actual opportunities you have to optimize your already existing content, how to expand your content to be provided as the answer by one of these devices.
If you have any question, if you have any doubt, please leave it in the comments, and I’m looking forward to seeing you in the next Crawling Mondays. Have a great week, bye-bye.