Disclaimer: This post, as with every post that aims to predict the future, should be taken with a grain of salt — no matter how authoritative the author.

The main purpose of this post is to offer ideas and open a constructive discussion around the future of SEO and digital marketing over the next 12 months.

2016 is, finally, close to its end.

It was an intense year, especially when it came to SEO and Google in particular. Because I'm deeply convinced that we cannot attempt any preview of the future without considering what happened in the past, I invite you to look back at the events that have marked the evolution of Google in the past 10 months.

It is important to note that, contrary to more classic Google timelines, I prefer to see all Google-related events in the same place. I believe it's the only way we can escape from a too-narrow vision of where Google is headed:

Click to view a larger image

Blue: Official Google Updates
Red: Businesses/companies acquired by Google/Alphabet
Green: Main posts in Google Webmaster Blog
Purple: Main Google patents published
Brown: Products Google launched in the market

What we can learn from this timeline?

This:

1. Google is steadily moving to a mobile-only world. Mobile-first indexing seems like the inevitable consequence of a year (or more) almost exclusively dedicated to evangelizing and forcing a change of mindset from desktop to mobile.
2. Albeit links are still essential for rankings (see Penguin 4.0), Google's investigative efforts seem almost fully devoted to entity, predictive, and personalized search. Again, quite logical if we consider deeply personal devices like mobile and home assistants.
3. For the same reason, voice search seems to be the next frontier of search, partly because Bing — using a different business strategy than Google — may represent a big competitor in this arena.
4. Since John Giannandrea has become the Senior Vice President of Search at Google, machine and deep learning began to be used by default in every facet of Google Search. Thus, we should expect them to be used even more in 2017, perhaps with specific algorithms improving Hummingbird at every phase.
5. In a mobile-only world, the relevance of local search is even higher. This seems to be the strategic reason both for an update like Possum and all the tests we see in local, and also of the acquisition of a company like Urban Engines, whose purpose is to analyze the "Internet of Moving Things."
6. The acquisition of startups like MoodStock and EyeFluence (but also Anvato and Famebit) seems to suggest that video/images and video/images marketing will be a central focus for Google, perhaps also because YouTube is struggling against Facebook (and not just Facebook) when it comes to videos/images and their monetization.

The shift from desktop-first to mobile-first

Until now, SEOs have considered mobile search to be one of the many specializations of SEO, on the same level as local search or international SEO.

That mentality did not change much when, back in 2015, Google announced AMP. Moreover, us SEOs considered AMP just another (often annoying) "added task" to our implementation checklist, and not as a signal of the real intentions of Google: Mobile search is all search.

With the announcement of mobile-first indexing, though, these intentions are now 100% clear, and somehow they represent a Copernican Revolution: After 18 years of prioritizing desktop, now we have to prioritize mobile.

The reason for this epochal change is evident if we look at the source of the search traffic (both organic and paid) for our sites:

Click to view a larger image

I designed this chart using the search traffic data Similarweb offers us. For all the industries categorized by Similarweb, I took the first five websites per search traffic volume in the USA during last November, and saw for each one of them how much traffic was from desktop and how much from mobile during the past three months.

Even though this analysis cannot be considered exhaustive and granular, as I hadn't considered the industries subcategories and I hadn't considered the "long-tail websites," surely it's indicative of a trend.

The results are clearly telling us that mobile search is bringing more traffic to websites than desktop: 20 industry niches out of 24 see mobile as their first source of traffic.

The four industry niche exceptions to this general rule are important ones, though:

1. Computer & Electronics
2. Internet & Telecom
3. Science
4. Travel

A good example of a website that still sees desktop search as its main source of search traffic is Tripadvisor.com:

• Desktop search traffic represents 71% of all traffic from desktop
• Mobile search traffic represents "only" 55.79% of all traffic from mobile

However, these same percentages should also make us reflect. They don't mean that TripAdvisor isn't visited on mobile, but that other channels are relevant traffic sources on mobile more than desktop (such as direct, not to mention the mobile-only app).

AMP, then, was the main character in the Google Search-branded storytelling about mobile this year.

Google announced AMP in October 2015, and by April already 37% of news sites' articles had an AMP version, according to a study by the GDELT Project.

However, the same study reported that, globally, only 40% of all news sites articles had a mobile version of any kind.

It must be underlined that the GDELT Project study refers only to news sites and not ecommerce or other kinds of websites, which see heavier use of mobile or responsive versions. Nevertheless, it can still be considered a good barometer of the reality of the web overall.

Speaking of "barometers," the Consumer Barometer with Google for 2016 is showing us important trends for the USA, like this one:

The percentage of people mainly using a smartphone is growing, while the percentage of people mainly using desktop is decreasing with respect to 2015 (or is stable if we consider the last 5 years).

Beware, though: If you analyze the trends in other countries, like some Asian or European ones, the percent of people using mainly smartphones is even greater.

Does this mean that we should neglect desktop search? No! If wid, it would be a big mistake, especially if our website were an ecommerce site.

The chart below, based on the same Consumer Barometer with Google data, tells us clearly that desktop is still by far the most-used device for product research (desktop is in orange):

Click to view a larger image

This insight must be considered if we're planning to redesign our site, to find a balance in terms of site usability for both desktop and mobile... and I cannot help but think that the subtle (and recently not-so-subtle) suggestion from Google of moving from mobile/responsive to PWA is also influenced by this reality.

What to plan for 2017?

Prepare for mobile-first indexing

When Google announced mobile-first indexing last November 4th, it did not say that the change would happen that same day, or even after a few days.

Google, instead, said this:

To make our results more useful, we've begun experiments to make our index mobile-first.

This means that we are still in a desktop-first index, but it's almost sure that it'll switch to mobile-first in 2017.

As happened with Mobilegeddon in 2015, Google is giving us plenty of time for:

1. Creating a mobile version with any possible format (m. site, responsive, adaptive, PWA) of our site if we still haven't (remember how few news sites' articles have a mobile version?).
2. Making the content and pages presented both in mobile and desktop versions the same. Be aware that this is the only possible way to really lose rankings, because if in desktop search we have visible content and pages that were discarded in our mobile version, when mobile-first deploys, it will lose that SEO visibility. For this reason, Google suggests responsive as the easiest way to avoid this problem.
3. Implementing structured data in our mobile versions, because it's usually neglected in the interest of speed (and Google needs that information!).
4. Eventually — and hopefully — reconsidering all the user experience and conversion optimization we offer on desktop and mobile (check out this deck by Talia Wolf from MozCon). For instance, in recent months — because of the Google demotion of tabbed content — many websites started to get rid of tabs and present all their content at once. This limitation won't apply anymore once mobile-first comes.
5. Rethinking and planning a new link building strategy if we have a separate m. mobile site. This is more of a defensive strategy suggestion, though, because we still don't know what will happen to inbound links to desktop versions in a mobile-first indexing world. It may happen that Google will find a way to make the Link Graph independent from the nature of the sites.

In light of what Google has told us about mobile-first indexing, and that you can find finely discussed here in this Q&A on Search Engine Land, If I had to give an extreme suggestion, it would be this:

if you have a very bad mobile version, and if you know that you're not going to have a new, fully functional one in time for the end of 2017, then (absurdly) it could be better for you to have a desktop-only site.

In fact, Google has repeatedly said that mobile-first does not mean that it won't index the desktop version of a site. To the contrary: If a site doesn't have any mobile version, Google will index and consider for rankings its desktop one. And this will be the case even if that same website has an AMP version.

Finally, I strongly urge you to update (or download, if you still don't use it) Screaming Frog.

In its very recent 7.0 version, Screaming Frog allows us to fetch and render crawled pages, something that before was only possible (and with a painful one-by-one URL process) via Google Search Console. Obviously, remember to set up Screaming Frog to emulate the Googlebot smartphone crawler.

Moreover, Screaming Frog now also alerts us to any blocked resource that could impede the correct rendering of our pages, again just as GSC does — but without the pain.

AMP

Despite some concerns AMP is generating amongst some bigs news sites, web owners, and SEOs, it doesn't seem that Google will reduce pressure for a large number of websites to adopt it.

On the contrary! In fact, if AMP was at first directed to news websites (and blogs), recently Google started presenting AMP results for recipe sites too:

And for an ecommerce website like Ebay (one of the founders of the AMP Project):

Therefore, if your website is already receiving a great volume of traffic from mobile search, you might start scheduling the creation of an AMP version.

This should be a priority for a blog, a news site, or a recipe site.

However, if you have an ecommerce site, it could be interesting to AMP-lify a category to test the performance and ROI of creating an AMP version of it, as the AMP Project suggests here. Not every functionality that's standard in ecommerce is possible with AMP, but if I had to bet, this is the niche where the AMP Project will see its biggest enhancements; Google and Ebay are too deeply involved to ignore it.

That said, if you are an ecommerce site, while it can be exciting to experiment with AMP, your real strategic choice should be going PWA.

Resources about AMP

• What You Need to Know About Accelerated Mobile Pages (AMPs) - Whiteboard Friday - Moz
• How Google's AMP Will Influence Your Online Marketing - Moz
• Should You Implement AMP Pages on Your Site? - Builtvisible
• Accelerated Mobile Pages (AMP) – What You Need to Know - TheSEMpost
• AMP Validation and Troubleshooting - AMPProject.org
• How to implement Google AMP on your WordPress site as easily as possible - Search Engine Watch
• wbAMP: Accelerated mobile pages for Joomla
• AMP for Drupal 7 and 8

Progressive Web Apps (PWA)

I am quite confident that If there's a main trending topic for 2017, it will be Progressive Web Apps.

Not only has Google already started evangelizing it publicly via its Webmaster blog and developer website, but Googlers are informally suggesting it in conferences and private chats.

As we've seen above, ecommerce websites are not yet fully AMP-lifiable.

Moreover, three seconds is the new fast, according to this study Google presented last September. Even a very well-optimized responsive or m. site can barely perform with an average SiteSpeed like that if we consider how heavy web pages are right now.

Then comes the other obsession of Google: security... and PWA only works with HTTPS.

So, it's as easy as summing 1+1 to foresee how Google will push websites' owners to go PWA.

The only setback to this evangelization, ironically, could be mobile-first indexing, which is still very uncertain in all its details, hence causing people to hold off.

However, if you're an ecommerce site, don't have an app, or are reconsidering the opportunity of constantly maintaining two apps (iOS and Android) because of the need to rationalize costs, then Progressive Web Apps can be your best choice, as they allow a website to work as if it were an app (and offline, too).

Again, as we sometimes forget, SEO's future will be determined on a macro- and micro-scale by business decisions.

Resources about PWA

• Introducing Progressive Web Apps: What They Might Mean for Your Website and SEO - Moz
• A Beginner’s Guide To Progressive Web Apps - Smashing Magazine
• Intro to Progressive Web Apps by Google on Udacity

Understanding language is the holy grail of machine learning

This phrase is the headline of the Natural Language Understanding Team page page on Google's Research website. The author of this phrase? John Giannandrea, Senior Vice President for Search at Google.

On that same page, we also find this:

Recent research interests of the Google NLU team include syntax, discourse, conversation, multilingual modeling, sentiment analysis, question answering, summarization, and generally building better learners using labeled and unlabeled data, state-of-the-art modeling, and indirect supervision.

With that declaration in mind, we have a better understanding of what Google is doing by simply looking at some patents this team has published:

• Annotating Topic Development in Information Seeking Queries
• Contextual LSTM: A Step towards Hierarchical Language Modeling
  
• Conversational Contextual Cues: The Case of Personalization and History for Response Ranking
  
• Transforming Dependency Structures to Logical Forms for Semantic Parsing
  
• Learning N-gram Language Models from Uncertain Data
• et al (see them all here)

Context, as you can see, tends to be recurrent, and — as anyone who's studied linguistic and semantics knows — this is a very easy thing to understand.

A classic example is how "carro" means "car" in Mexico and means "carriage" or "wagon" in Spain. The meaning of a world can radically change because of context, in this case cultural context.

Context is fundamental for understanding the meaning of the implicit and compound facets of any conversation, which is fundamental for the successful development of completely new search environments like Google Assistant and Google Home. Will Critchlow (with the collaboration of Tom Anthony) explained this well at The Inbounder last May:

Finally, context and natural language are partly the basis (for what we know) of the infamous RankBrain, as vectors too are contextual (and contexts by themselves), as Bill Slawski explains in this post.

Moreover, Google finally seems serious about understanding one of the most common (and most complex) aspects of natural language: metaphors. And once they're able to understand the meaning of metaphors, understanding the meaning of all the other rhetorical figures people use when talking (and writing) will be an easy incremental step for Google.

Why does Google have this irresistible interest in natural language?

Sure, on an ideal level, it's because Google wants "to provide the better answer to users' needs," and to do that, Google must:

1. Understand what each web document is about (semantics);
2. Understand what users are actually searching for, now more and more using their voice and typing in the search box (natural language processing).

Another ideal reason is that "You don’t need to be at your desk to need an answer".

But then there are more earthly (aka: business) reasons:

1. Voice Assistants adoption is growing, as reported by MindMeld;
2. 40% of Voice Assistant users started using it only 6 months ago, which is a sign that the "early adopters" phase seems passing testimony to the "mainstream users" phase;
3. Already 20% of mobile queries are Voice Search (and will be 50% in 2020, according to Microsoft);
4. The primary setting for voice search is our home (followed by our car), according to a KPCB study;
5. Already in 2014, Google was reporting that the biggest percentage of voice search users were teens. Those teens are growing, getting jobs, and becoming parents;
6. Amazon sold about 5 million Amazon Echo in the last two years, and Amazon reported that "Echo Dot and Echo Tap, two smaller and cheaper versions of the traditional Echo device, accounted for at least 33 percent of sales in the past six months" (source: Geekwire).

These are some of the reasons why Google developed Google Assistant (a completely new search environment, as defined by Google itself at Google I/O) and Google Home, and that's why Google announced that Voice Search reporting will come to Google Search Console in the future.

As in the past with Android, Google is entering in the market when all its competitors are already established. Let's see, thanks to the adoption of Android as a mobile OS, if it will be able to recover its position and eventually become the market leader.

What to plan for 2017?

When it comes to natural language, voice search, and intelligent voice assistants, what SEOs can do is more related to optimizing for the parsing and indexing phases of the Google algorithm than to rankings themselves.

I talked about this idea in my last post here on Moz — Wake Up, SEOs – the NEW New Google is Here — so I won't repeat myself, but I do suggest you go read it (again).

More concretely, I would plan these tasks for the first month of 2017:

• Implement structured data in the mobile version of my website.

This is needed not only because Google will eventually use mobile-first indexing and these usually don't have schema implemented, but also because structured data is one of the fundamental tools Google uses for understanding the meaning of a web document. Moreover, Google is really pushing rich cards for mobile search, somehow replicating the incentivization strategy used in the past with rich snippets.

Are you worried about performance? Then it's time to adopt JSON-LD (paranoiac thought: is this also a reason why Google strongly insisted on JSON-LD for structured data?).

• Optimize for #0

Featured snippets are even more prominent on mobile search, and also used by Google Home to offer answers, even though — as wisely said by Dr. Pete here on Moz — it's still not clear how that will translate into a click to our website.

Right now there are several tools that allow us to investigate and know what queries fire up a featured snippet (apart from all the other SERPs features). The ones I use are these:

• GetStat
• Moz, both Keyword Explorer (SERPS Analysis) and Moz Pro Campaigns (SERP Features and Analyze Keywords in the Rankings section)
• SEMRush, in the Positions page of the Organic Research section of its Domain Analysis, which also offers to visualize a snapshot of the SERP for each keyword

SEMRush SERP snapshot, and an example of how a featured snippet can be generated from an ecommerce category page (the SEO for Amazon must be very happy, I bet).

• Start using the Google Assistant API and experiment with custom voice commands
• RankBrain is one of the fundamental bricks toward a natural language based search engine, so if you have not already done it, start rethinking keyword research, and stop generally talking about "topics" with no real actionable strategy behind it.
• Consider branding as an SEO strategy

One of main characteristics of Google, enhanced by entity search and context, is personalized search.

Personalization, then, seems to be even more important if we consider personal assistants.

Personalization means that Google will more often present content from websites that are in our search history or — through search entities — linked from websites already present in our search history.

This means that if, on a short, tactical level it's important to target long-tail queries, on a longer, strategic level the ideal is making our brand synonymous with our products and services. This can be achieved by targeting higher up the funnel with the right content in the right format, published and promoted at the right moment to the right people. This is very well described Jono Alderson at Searchlove London in 2015 (here's the video recording):

Searching higher up the funnel from Jono Alderson

• Reconsider Bing!

If you think that Bing is only "that search engine with cool background photos," it's time to change your mind. Bing is fueling the search of Siri and Alexa, apart from being the default search engine of Cortana. If you calculate how many iOS/OS X, Windows 10, and Amazon Echo devices are used, then you have a rough idea of how Bing could be important as voice search grows. You can read more about voice search, Cortana, and Bing in this post by Purna Virji.

SE.LO.MO (Search Local Mobile)

Only a few years ago digital marketers used to talk a lot about SO.LO.MO. (SOcial, LOcal, and MObile).

We were all talking about Foursquare marketing. Then Foursquare changed to Swarm, and we no longer talk about SO.LO.MO., partly because the marketing philosophy behind it has become a default practice.

However, now with mobile as the first search traffic source and the unstoppable success of personal assistants and chatbots (I invite you to look this deck by Jes Stiles), the idea of doing marketing locally and mobile is even more pressing and promising, although there are technologies like beacons that don't seem able to conquer the market, maybe because they're too advanced with respect to consumer behavior.

Returning to TripAdvisor as an example, if we look at which queries bring more organic traffic from mobile (and excluding branded searches), we see these:

Click on the image to see a larger version

Apart from telling us that people in the USA really like to go out for dinner without any clue on where to go, what this simple analysis shows is that people search on the go more and more. They've finally understood that searches are localized and they don't need to explicitly indicate their location. Perhaps even more important, they now fully know that their results pages are personalized.

What this snapshot above is not telling us, though, is a trend that could become the new normal in the next future: longer verbose queries because of voice search.

In fact, if we dig into the Similarweb mobile keyword report, we can start finding these kind of queries:

• hotels on highway 41 near grand ave and rogers ct near waukegan il;
• places in clarksville arkansas that fix cracked windshields;
• restaurants open late in west side of south bend indiana;
• five star restaurants like salvatores in western new york;
  
• where to get breakfast near grand hotel francais paris.

The SERPs answering these queries, though, also show us one problem and one opportunity:

1. The problem is that these SERPs, while having a clear local intent, quite often do not present any local search pack.
2. The opportunity is that, despite these queries indeed being "local," Google fails to offer relevant results able to answer them (see the "five star restaurants like salvatores in western new york" as an example).

Therefore:

1. Thinking of local search only as MyBusiness optimization may limit the opportunities businesses (especially local businesses) can have to earn SEO visibility and traffic.
2. Also, local business websites should start working to intercept the potential traffic generated by those kinds of queries. There is a real opportunity in those kind of queries, simply because (still) nobody is really thinking about them (apart fromTripAdvisor, as its result for "where to get breakfast near grand hotel francais paris" testifies).

How to achieve that?

Probably not by trying to target all the infinite possible combinations of local searches a user can do in relation to our kind of business and our location. That would be equal to creating content of very poor quality, when thinking about how Hummingbird and, in many aspects, RankBrain work.

This leads me to redirect you back to the previous chapter of this post about semantics, natural language, and context.

David Mihm advises to Think of your website (or your client's websites) as an API, adding that:

Even if you’re not a publisher in the traditional sense of the word, you should prepare for a time when no one ever visits your website. Awareness, research, and conversion will all happen in search results, and the companies whose websites facilitate that paradigm on the leading edge will be rewarded with more customers while competitors scramble to catch up. This means as much Schema.org and JSON-LD markup as possible, and partnering with third parties that have cut deals with Google to facilitate transactions (see: OpenTable and ZocDoc).

Because David is surely more expert in local search than I, if you want to dig into what could be the trends in this very important SEO area, I invite you to read his predictions in the Tidings blog.

THE IRRESISTIBLE ASCENT OF VIDEO (and the images strike back)

Video

This post by Contently is old (July 2015), but it shared still-interesting stats via an Emarketer study about the growth of video consumption online.

Consumption of video online is growing, even though — apparently — it's not really stealing time away from TV.

Things look different if we look at "generations" (I don't really like marketing segmentations such as "millennials," but we can still use them for brevity):

What this chart is forgetting, though, is the youngest audience (from 4–13 years old). If you have a child around those ages, you'll agree that she consumes video mostly online. For instance, my kids' idols are Iron Man, Aragorn, Luke Skywalker, and DanTDM, a YouTuber, who shares his videos while playing games like Minecraft.

Let's add a final stat about what device is used the most for watching videos online:

Laptop & desktop are still the most-used devices, but smartphone is quickly growing.

Consider that:

1. The average age kids start owning a smartphone is 10.3 years;
2. Children from 5 to 13 years old (and also young people up to 20 years old) tend to me more visual than textual;
3. Their influence on the buying habits of their parents has been known for many years and, in 2012, it was equal to $1.2 trillion USD in spending.

I talk a lot about kids because they are the most crystal-clear example of why every technological platform is so devoted to video and video live streaming right now (Instagram being the last one announcing it on 11/21). However, this trend in consuming videos online (and YouTube still is the most-used channel) is common to almost every age group of Internet users.

Finally, if we pair to this video frenzy the equally irresistible rise of native advertising (pro tip: follow Melanie Deziel), then see that Google acquired companies like Anvato (a "video platform that guarantees video playback and monetization from signal to every screen" as it describes itself) and Famebit (an "Influencer Marketing Platform for YouTube, Instagram, and More", as its title tag recites) is not a surprise at all.

Google needs to find new ways of monetizing videos... and YouTube is not enough anymore.

More concretely, if we think about our 2017 SEO and digital marketing strategy, video seems to be a channel that we should start exploring more seriously, if we did not consider it before.

And when it comes to digital PR, we should start considering online videos stars as much (if not more) influencers as any classic blogger.

Images

In 2017, for almost the same reasons explained above for video, we should expect a return in interest for images marketing, especially in Google search.

Let's be honest: Images search, as it is right now, is the dinosaur of Google search. For us SEOs it hasn't been useful in bringing traffic to our websites for many years and, for Google, it's not profitable.

Maybe this is one of reasons why Google bought Moodstock and invests so much intellectual and machine learning efforts in image recognition.

People do showrooming... They go into a store, take photos of products with their smartphone, and then search for online offers for those same products.

It should not be silly to think that Google could "help" this search thanks to image recognition, because it already does it quite well with its reverse image search feature.

Moreover, with Schema.org/Product, we can tag the images of our products so that Google can easily pair product images to other characteristics like prices, offers, and stock availability.

With this data, it could start monetizing the Images vertical once for all.

[NOTE: As I was writing this, Google announced that it will start showing product schema rich snippets in image search results... so this is no longer a risky preview, but partly a reality!]

VR

What is the last company did Google acquire? Eyefluence.

What physical product did Google launch in October? The DayDream VR headset.

What was the most exciting feature that YouTube (and other platforms) rolled out? 360.

I'll let you fantasize about the opportunities VR represents for the smart digital marketers.

Maybe in 2017 VR will still be an "early adopters" technology, but if I were you, I'd start preparing myself and clients to it.

---

Credits: The images in this post, if you didn't guess it already, are from the HBO show Westworld.

The captions in the photos are from sci-fi movies and TV series titles (have fun discovering them)