There has been a lot of discussion in the past few years about SEO voice search and how it has been changing SEO in many ways. I tend to agree. I also suggest that the acoustic nature of voice search will hasten the development of new ways to present your website data specifically targeted to those using voice search devices.
We’ve seen plenty of online articles about voice search marketing that refer to the necessity of preparing for voice. Many reference little more than the importance of long tail keywords and lengthier query structures associated with voice search. This also means lengthier content more aptly zeroed in on the voice searcher.
While these might be part of an advantageous SEO strategy to optimize for voice, it can be very difficult to accurately and effectively annotate for mobile search intent and voice search intent within a single website page.
In this article, we will explore several SEO voice search considerations.
Keyword Text Content vs Voice Content
Some SEO voice search strategies may work for a while, but voice searchers are looking for more than just text content. They’re looking for voice content. And as Amine Bentahar wrote in his Forbes article 2017 Will Be The Year Of Voice Search:
“No longer can you simply target a specific keyword, but content and websites need to be more complex, ready to answer the questions that your users have.”
The complexity of SEO voice search is still in its infancy, and research has presented us with some interesting tangents to follow. In a paper titled Understanding User Satisfaction with Intelligent Assistants (from March of 2016), Julia Kiseleva et al. studied “how user satisfaction varied with different usage scenarios and what signals can be used for modeling satisfaction in different scenarios.”
We know that Google is all about searcher satisfaction, and they aren’t afraid to reduce the rankings of sites not keeping up with the rest of the flock. Check out our earlier article about Google’s interstitial ads penalty update as just one example of their prioritized commitment to the searcher. And, rightfully so. It is the needs of the searcher on which Google’s empire has flourished.
Voice Search Behaviour is Complex
Julia Kiseleva et al went on to write:
“We also study how the nature and complexity of the task at hand affect user satisfaction and found that preserving the conversation context is essential and that overall task-level satisfaction cannot be reduced to query-level satisfaction alone.”
Starting to see what we mean about complexity?
Granted, their voice search research is on intelligent personal assistants, but the searching via these assistants for in-depth queries through online channels are in many ways what will push the edges of emerging dialogue surrounding how voice searches might be optimized.
Keywords for SEO Voice Search Differ from Text Search Keywords
Add to all of this the fact that voice searches and text searches are very different.
At the moment, there’s a lot of theory, conjecture, and hyperbole about SEO voice search, but little clear direction. In the Forbes article, Bentahar also indicated that “[k]eyword research is needed for how people like to phrase questions out loud, understanding the search terms consumers are likely to use.”
This in and of itself opens an enormous can of regionally-specific, colloquial worms. I can’t help think of the off-kilter, axiom-like joke that Canadians sometimes pass around at parties. It goes something like this. Canadians ask “How’re you doin’, eh?” Someone from New York might ask “Hey, how you doin’?”
Voice Search Behavior and Good Abandonment
As Julia Kiselev et al. wrote in their paper:
“We know that user satisfaction for mobile web search is already very different. So we cannot assume that users who do not interact with the SERP are dissatisfied. This problem of ‘good’ abandonment received a lot of interest in recent years.”
I have found the same. I read a couple of papers on good abandonment including Good Abandonment in Mobile and PC Internet Search, and Detecting Good Abandonment in Mobile Search. I came to the conclusion that good abandonment (the act of nothing after mobile search results are displayed to the mobile searcher) can’t possibly be construed as much of anything. The absence of something (searcher action post-query), does not indicate the presence of something else, (searcher satisfaction).
I think that the search engine research community have aligned themselves so closely with good abandonment because frankly, there are yet no other identifiable means of determining what an abandoned search means. Stop and think about this logic …
Picture yourself sitting in a doctor’s office waiting to be called into the examination room. You pull out your phone and enter a search into Google for a company name that you saw on the side of a vehicle on your way to the doctor’s office. Just as the SERP finishes loading, you’re called into the examination room. You put your phone to sleep, but the browser is still open. You forget about it until you are about half way home again. Is that too considered good abandonment?
There are dozens of similar scenarios where good abandonment makes no sense whatsoever. You can’t assume something about voice abandonment any more than you should associate lack of interaction with the browser post-text-search as good abandonment. The phrase good abandonment is at best a term used because there is no other data on which to quantify abandoned search results.
Additional Usage and Statistics
Voice Search Queries Still Far From Natural Language
Fortunately, researchers like Ido Guy can give us some ideas of what to focus on. Guy is a Principal Research Engineer at Yahoo Research. In his paper Searching by Talking: Analysis of Voice Queries on Mobile Web Search, Guy shares, “we (presumably Yahoo!) perform a query log analysis of half a million voice queries, issued to the mobile application of a commercial web search engine, over a period of six months.”
Not only was the voice query log analyzed, it was compared to sample text queries on mobile of the same relative size.
Guy further reports that:
“…(w)e provide empirical evidence, based on language modeling, that voice queries are closer to natural language than text queries, yet are still distant from natural question language.”
The analysis was based on 500,000 random voice queries performed by 50,000 unique searchers, over a period of 6 months in 2015. Voice queries were predominantly more frequent during the day from 8am to 8pm, and text queries were more frequently carried out from 8pm to 8am. These stats were consistent 7 days a week.
Voice query lengths of 5 words or more comprised 34.5% of all voice queries, as compared to only 21.2% of text queries. On average, voice queries are definitely longer than voice. But what I find most interesting is the distinctive term set used by voice as opposed to what is used during text queries.
Terms most used on the voice list include:
- Question words
- Function words like determiners
- But rarely any nouns.
On the other hand, the text search list included numerous abbreviations for states, such as nc, tx, ca, etc. These were almost never used on voice searches.
Voice queries usually began with question words How and What. The most distinctive words used on a text search were Facebook and Pornhub.
Guy also wrote that, “A recent survey of 1400 U.S. smartphone users found that 55% of the teenagers use voice search every day.”
In an article titled How Voice Search Will Change Digital Marketing – for the Better, Purna Virji refers to a voice search study by Thrive Analytics that indicates 71% of mobile personal assistants users are 18-19, and 59% are 30 to 43.
But even with widespread usage of voice for this particular age group, click-through rates are substantially higher for text searches than for voice searches, and at greater than 2:1 ratio. Not surprisingly, voice searches focused more on audio/video content. And that is where the most dramatic rift between text and voice search begins to occur.
Trends and Behaviour
Overall, voice searches also tended to focus on topics that required less screen interaction than text searches. This is another clue for the SEO community that voice is undoubtedly evolving into its own, distinct taxonomic search group.
Guy goes on to write in his study that his “…findings suggest that voice queries pose their own type of language, in-between traditional text queries and natural-language questions,” and that, “…new metrics for evaluating user satisfaction of voice queries should be developed.”
These glimpses of a new taxonomy of searcher, slowly being revealed through research paper after research paper, is beginning to outline how differently everyone will have to examine SEO voice search.
Voice searchers use distinctly different queries than text searchers on mobile. And even though there may be similar, overlying characteristics, voice searchers are not using natural-language questions per se.
The Future of SEO Voice Search
Voice search will require a whole new set of measurement devices, an entirely new lexicon of terminology, and a willingness by SEO strategists to explore all avenues of presentation to voice search users. With so many details still to be examined, and so many questions to be explored, SEO voice search is certainly going to be interesting for the future of search engine optimization.
If you have any questions about SEO voice search, how to develop a voice search strategy that makes sense for the shifting landscape of vocal search, or you just want to contribute your two cents worth to the discussion, contact us today!
Email: [email protected]