Voice Assistants Are Changing Content And SEO

Speech is rapidly becoming the new UI. Instead of clicking and typing, we’re slowly starting to talk more to devices. It is faster, more convenient, it’ll be all around us, and it’s the more natural UI. The way we search will change dramatically. And with it, the way we present content and how we perform SEO.

As smart home devices such as Google Home, Amazon Alexa, Apple HomePod are becoming ubiquitous, the way we search – by speaking – has an impact on Search Engine Optimization (SEO). These devices change our queries, as we use natural language to perform searches, instead of phrases and keywords we type into a bar. We’ll ask our device questions, like “What is the best place to grab lunch around here” instead of “best lunch [city]”, for example. Furthermore, the virtual assistant uses more contextual information (who is asking, why, when, in what thread, et cetera) than the old fashioned search engine did. 

Voice is king

With advancements in Natural Langue Processing and subsequent rise of conversational AI, the technology for a veritable seachange has been in place for a few years now. What we’ve seen in the past, is that technology is only one part of the equation: culture is often even more important. The desire to use new gadgets and new methods is what ultimately drives technological change. So will we end up adopting this available technology? All signs point to “yes” as devices like Amazon Echo and Google Home are selling better each year, and more people turn to assistants like Siri or Alexa to quickly find the answer to what the weather is going to be like today in their area, instead of typing something like “weather [city] [today]”. According to data from Statista, the number of voice assistants in use worldwide will grow to 8 billion by 2023.

What is also surprising, is that it’s not the usual demographic of early adopters, 20-somethings, that turn to voice search. Research shows a relatively high number (relative to usual technology adoption figures) of 40- en 50-somethings that turn to this newer user interface. That would indicate that it is indeed the more natural interface, as people who didn’t grow up with touch-interface, seem to switch to voice relatively easy compared to switching over from type to touch. That is a fascinating development for UX design and illustrates the importance of voice search during the coming years.

Change is ongoing

This evolution has been a long time coming. The renewed interest in Machine Learning and Natural Language Processing in the past decade had developers excited for the future of voice. Indeed, at Microsofts developer conference Build in 2016, CEO Satya Nadella was optimistic about the era of even more personal computing, how this would change processes, and drive a need for more intelligent platforms. Data input is becoming more and more natural: it moved from type to touch to voice. Voice is the logical conclusion of a digital computing journey we’ve been on for over 75 years.

Smart speakers change the input of searches from simple text to something more detailed, and it’ll also look for specific answers rather than query results themselves. A result stating, “This is the best [place] in [city]” will score, all things being equal, higher than something like, “What is the best [place] in [city]?” – although an important caveat is that Google will certainly massage its algorithm to fit the content available at large. Meaning, this is a theoretical example; in practice, all the headlines ending in question marks will still score big, as much of the organic traffic is still being directed there. But as we move away from text, so will organic results move away from these structures, changing the way we present content online, and how we optimize for it.

The question is, “How much are things changing?” For one thing, the results page will be much less important as, say, a featured snippet. A voice interface searches for an answer, not the possible results of a query. It will tend to focus on and read snippets, not direct users to a webpage. It feels like a failure when virtual assistants come back with a reply like, “This is what I found on the web”. That’s not a relevant answer to the question. 

Increased interest in snippets and FAQs

As a purveyor of content online, you don’t just want to score high on SERP anymore, you’ll want to score number one (preferably on a contextually rich question) or have a featured result. Content mainly focusing on getting one question right, will, therefore, be rewarded over content that tries to explain or answer multiple questions. Current research – for instance, this study by Backlinko – shows that Google Home focuses on short, concise answers, as opposed to deeper, meaningful content.

That is, insofar as Machine Learning used in the search algorithms isn’t evolved yet enough to parse large texts to look for specifics within results. As such models do already exist, it is only a matter of time before ML and AI used by search companies become smart enough to answer questions without looking at specific directions. So for the coming months, perhaps even years, you can expect to see more FAQ based and to-the-point snippets of text. But after an initial period, meaningful, informative texts will become the content that search engines will focus on as they develop methods to extract answers from those texts.

In the meantime, featured snippets will become even more sought-after, so expect content to fit that mold. Sites looking for traffic would much rather focus on a hyper-relevant answer, than on a deeper explanation as to why. More than ever, detailed background information seems destined, for a few years at least, to become a niche for those who are interesting in specific content – as most users are not looking for the ‘how’, but the ‘when’ and ‘where’ – while directions to the answers of frequently asked questions will become more relevant to search engines. 

Specific actions for SEO

If this is indeed the case, organic traffic will lead logically to subpages with more specific answers, while direct traffic is the most important vector for creators of highly specialized content. So, in all, it is not much different than the last decade; the big change is that generalized content is likely to fall on the wayside. To make content stand out, more than before SEO specialists will need to look for pages that score well in analytics, but also are less common among competitors. Tools like Google Search Console, Sistrix, SEMRush, etcetera, should help in the analysis of SERP position relative to competitors. 

Because the future of search is more mobile than ever, page load times are increasingly important. It doesn’t help a user to stand around for a few extra seconds while Alexa loads in a response based on a slow loading page, so it follows that Google tweaks its algorithm to even further downrank slow websites and promote fast loads. This has been a trend for many years now, and will only become more prevalent with the rise of smart speakers. Indeed, the Google Home study by Backlinko, referenced earlier in this article, indicates that the average voice SERP load times are 52% faster than the average page. 

What’s next?

As voice search has become a priority for companies like Microsoft, Amazon, and Google, it will undoubtedly evolve faster than search based on type, simply because of the greater amount of resources dedicated to it. It is just a matter of time before machine learning models are implemented to query unstructured data, longer texts, and elaborate multimedia pages to find specific answers: the era of FAQ content is likely to be just a band-aid in the meantime.  

Furthermore, contextually based answers are becoming even more important. Consider someone strolling around in Amsterdam, but living in Utrecht. When she asks “where is the nearest HEMA”, a contextual approach can be as simple as using location data to ascertain the nearest store in her vicinity. But, if she were to converse with her assistant, saying “remind me to go to the nearest HEMA on Saturday”, the context changes and the assistant will make a reminder for Saturday, conclude that it is likely that this will be in Utrecht, look for stores there and add these to the calendar appointment. And that’s just the simple approach. A trained assistant will anticipate why the user wants to visit a HEMA, which location she is likely to travel to, and what alternatives there are to get her desired item, for instance from a store that is closer by or a store that sells the item at a lower price. 

In short, voice search will look at a combination of intents (what does the user want) and entities (where, when, etc) to tweak its results in a way that suits the user more than a generalized answer to the question. Expect SEO to match this change, with a stronger focus on conversational answers than keywords. That is an interesting departure from the previous two decades.