Thursday, June 23, 2016

Why retailers shouldn’t overreact to the voice search revolution

If recent accounts on the rise of voice search are anything to go by, the volume of long-tailed queries with more natural language and searches with a question is heading nowhere but up and to the right. This, the argument goes, should in turn impact our digital strategy as we strive to account for the inherent differences in typed search vs. voice search.

Taking a look at the search queries triggering paid and organic results for retail brands using Google’s paid and organic reporting in AdWords, however, there hasn’t been much movement over the past couple of years for a few key query attributes that would indicate a major shift in search behavior.

This makes the excitement surrounding voice search sound a lot more like those way-too-early “year of mobile” declarations than anything that needs to be rapidly addressed by all sites and brands.

And while the research presented here is far from the be-all, end-all in terms of measuring the impact of voice search, we posit that even if voice search does take off and change the types of queries searched for, the resulting best practices look almost identical to our existing paid search and SEO best practices.

But before I go on, BIG thanks to Merkle SEO Director Jamey Barlow for contributing the SEO thought leadership for this piece.

Little change in overall query length over the past two years

Microsoft research analyzing Cortana query data shows that users are more likely to search for longer queries when speaking searches than when typing them (though we’ll speak later about why this analysis might be a little misleading).

cortana_query_word_count

If voice search were really taking off in a meaningful way, then, we should expect to see a greater share of searches attributed to longer queries over time, according to this research.

Taking a look at the character length of queries that trigger paid and/or organic links for Merkle advertisers based on Google’s paid and organic report in AdWords, we find almost no steady trends since March of 2014 to suggest that queries are getting longer.

paidorganic_query_character_length

Character count is only one way of assessing how long-tail a query is, though. What about word count?

Again, no steady trends to indicate word counts are going up:

paidorganic_query_word_count

If voice search is on the rise, why no increases in word count or character length?

There are a few possibilities to mention here in explaining why we aren’t seeing longer queries:

  • Different industries might be more impacted. Maybe voice search isn’t a big deal right now for retailers specifically, which all of the brands studied are. I’ll speak more to why this is probably the case later, but fingers crossed on Google Search Console voice search reporting actually becoming a thing, so that we can track this traffic more confidently.
  • Comparing like queries is important. Comparing query length of voice search to typed search isn’t very meaningful right now, as people are currently only likely to use voice search for limited types of queries, compared to typed searches spanning a much wider array of query topics. The only way to make this analysis meaningful is to compare queries that are typed rather than voiced within the same topic — for example, comparing the length of typed vs. voiced direction queries. If voice searches are roughly the same length as typed searches when looking at only queries within the same topic, then overall query length across all searches should become more comparable between voice and typed search as users expand their use of voice search to more topics. One point to note here is that there are types of queries that will likely never be spoken, particularly in public, and it doesn’t take much imagination to think of topics this might apply to.
  • The paid and organic report isn’t a perfect data source. These are only queries contained in the paid and organic report. The fact that only about 75 to 80 percent of all paid search clicks are attributed to queries in this report indicates that it does not include all queries that trigger a paid and/or organic link. In turn, this almost certainly biases the results toward shorter queries. However, looking at the share of official paid clicks accounted for in the paid and organic report, there’s been no decline in this share YOY to point to more voice searches piling into the unaccounted for query bucket.
paidorganic_vs_official_adwords_clicks

Thus, if voice searches are indeed growing, we would expect them to be just as likely to get populated in the queries in the paid and organic report as typed queries.

These charts don’t disprove that users are increasingly using voice search, as there’s clearly research showing that users are warming up to this style of searching. However, they do show that retailers aren’t seeing much in the way of overarching change in the length of queries that are reported as driving paid and organic traffic.

Popular voice search queries have little connection to retail

Taking a look at a Northstar Research survey commissioned by Google, teens and adults differ slightly in the types of searches they conduct through voice search.

google_northstar_voice_research

However, the query topics for both age groups have very little connection to searches retailers would care about, except perhaps direction queries when it comes to brands with brick-and-mortar stores.

Shopping didn’t even make this list, either because the share of searchers who speak shopping queries is very low or Google didn’t find it important enough to include in the survey.

As such, there’s just not much reason for retail brands to significantly adjust their paid search or content strategies as a result of users becoming increasingly comfortable with voice search — yet.

Optimizing for voice search looks a lot like optimizing for search in general

Even if different types of queries do start creeping into search term reports from voice search, it’s hard to believe that brands would really need to adjust their strategy much; optimizing for a potential rise in voice search looks a lot like our existing best practices for both paid and organic.

Paid search

On the paid search side, advertisers should already be looking to launch keywords for those queries that produce meaningful traffic and to have processes for identifying these queries.

But again, it’s not totally obvious that voice searches differ much from typed searches, given that any research on the subject is comparing the small subset of topics users like to use voice search for to the wider variety of topics users type searches for.

Further, as we reported on in our Q1 Digital Marketing Report, 70 percent of e-commerce non-brand paid search traffic is now coming from Product Listing Ads.

rkg-q1-2016-paid-search-google-pla-click-share

Proper Google Shopping campaign management does require marketers to segment traffic to different PLA campaigns based on important elements of the query using negatives, such as the presence of particular product brand names in the query, as well as searches for each advertiser’s own brand name.

However, it’s unclear that voice search will result in any new query segments which most advertisers will have to account for in PLA campaign structure that aren’t already being accounted for.

Thus, advertisers should stay attentive to search query reports to look for potential new keywords and negatives, as well as Shopping Campaign optimizations, as they always have, but it’s unlikely that voice search is going to require much more.

Organic search

In 2013, Google rolled out a new version of their search engine algorithm named Hummingbird.

This update was another step forward for semantic search and focusing on better understanding user intent. Not satisfied to simply try to improve search results, Google was trying to improve your query and to add any helpful context they could around it.

The most obvious initial evidence of Hummingbird was in the synonyms it would literally highlight on SERPs. Where once, Google would put into bold type only the words that matched a user’s query, it now also showed the words it thought matched user intent.

For example, a search for “best lodging in yosemite” will also make bold the words “hotels” and “accommodations” as equivalents of “lodging,” and it also makes bold “luxury,” which it finds synonymous with “best” in this travel-related search.

yosemitelodging_screenshot

Given that Google could now better analyze each word in your query and derive a deeper meaning from the context created with those words, the implications of voice search seem rather muted. Your long, rambling voice search, filled with slang and conversational speech, is being exponentially expanded with synonyms to bring back pages that better fit your intent beyond matches for just a few keywords.

So we are back to ensuring that the content on your website matches user intent, because you can’t expect your pages to rank well in SERPs unless they answer a user’s question or help them complete some sort of task. Beyond making your website technically sound and fully crawlable by search engine spiders, that utility is the reason for your website and what should always have been your mission.

A potential rise in voice search does nothing to change this.

Despite this latest non-event, there are still several SEO optimizations to be done:

  • Ensure that you are addressing customer intent. Is the content on your site unique and of service to a user? A thorough competitive keyword and gap analysis is crucial to understanding what kinds of content to create and curate on your website. Can you leverage your internal site search data to further understand what users are searching for after entry?
  • Optimize for Quick Answers. Quick Answers is powered by the Google Knowledge Graph and often is returned on SERPs for informational queries, including “how-to” and “what is” queries. These Quick Answers usually appear as direct answers or featured snippets. The direct answers originate from Google’s knowledge base and answer simple queries. The featured snippets come from third-party sites like your own, and they provide short and descriptive answers, as well as links to your site. The commonalities for websites that appear in quick answers are authoritative pages with quality content that is thematically relevant and answers specific questions that match user queries. Google seems to prefer numbered lists and bulleted step-by-step instructions as well.
  • Optimize for local. Many voice searches are assumed to have local intent, whether the user is looking to order pizza from the corner restaurant or find the closest ATM to their location. We recommend optimizing your Google My Business listings, ensuring that your local pages have unique URLs and are indexed, and including name, address and phone numbers (NAP), as well as other relevant data in your on-page content that allows a user to find and get to your place during business hours and for special events.
  • Schema markup. Structured data helps spiders parse data within your website. We recommend using a JSON-LD script inside a script tag to make structured data available to search engines for your content.

Conclusion

Thus far, retailers have seen little in the way of change arising from users becoming more comfortable with voice search, at least in query reports.

This could be because the topics users are most likely to use voice search for don’t overlap much with queries of importance to retailers or because voice searches are really not that different from typed searches, despite arguments to the contrary.

However, even if a rise in voice search does begin to change the queries that drive paid and organic traffic for retailers, it isn’t obvious that our best practices will change much.

Time will tell, but for now, we recommend staying calm and continuing to optimize for paid and organic search using the same techniques that have worked since before voice search became the big news topic of the day.

The post Why retailers shouldn’t overreact to the voice search revolution appeared first on Search Engine Land.

No comments:

Post a Comment