Google Home, Pixel, and the Future of Search

NOTE: an edited version of this post has also been published on the company blog of Bruce Clay, Inc. as “How to Optimize for Google Home NOW”.

ICYMI, on October 4th, Google held a press conference that felt like their take on a muted Apple product launch; in lieu of the latter’s trademark dramatic blacks and sleek silvers, we had shades of bamboo and eggshell, and each presenter was dressed in that NorCal blend of comfy-but-costly hiking/laid-back-office attire; they all went by first names only, and amidst a smattering of slickly-presented data, launched several new products. (For a full run-down of the new products, you can check out https://madeby.google.com/.)

The star of the show was undoubtedly the new Google Pixel phone; a quick comparison of the new products’ search volume on Google Trends shows that interest in the phone far outstrips everything else, with nearly seven times the search volume of Google Home, their answer to Amazon’s Echo. 

Google Trends for Google Pixel and Google Home
Google Trends graph showing searches impacted by the announcement

For marketers and tech-watchers in general, however, it’s the Home that should be of particular interest. The entire time I was watching the announcement, I couldn’t shake this vague sense of change; that we are on the cusp of a real sea change in the way that we interact with technology. While Google Home certainly isn’t the first in-home speech activated device, the fact that Google is making one and launching it next to their new phone is a clear reminder of their interest in building a computer assistant taken right from classic sci-fi. Amit Singhal famously called the Star Trek computer “the ideal that we’re aiming to build – the ideal version done realistically.”

Voice Search and the Star Trek Computer

Scotty-Hello, Computer

I’d like to break that open a bit, actually; there’s an interesting episode where two characters have an unusually long interaction with their ship’s computer; in the episode “Darmok,” which was released in September 1991 (a month after “the first Web site,” http://info.cern.ch, went online, and 2 years before Google began), the ship’s counselor, Deanna Troi, and Commander Data, an android science officer, are trying to decode the speech of a newly-encountered alien race. Lacking any intergalactic Rosetta Stone, the two start their investigation by talking to the computer, asking if there are any previously recorded uses of individual words; in particular, an often-repeated noun: “Darmok.” The response they get back is surprisingly similar to the “10 blue links” of a SERP:

TROI: Computer, search for the term Darmok in all linguistic databases for this sector.
COMPUTER: Searching. Darmok is the name of a seventh dynasty emperor on Kanda Four. A mytho-historical hunter on Shantil Three. A colony on Malindi Seven. A frozen dessert on Tazna Five. A
TROI: Stop search. Computer, how many entries are there for Darmok?
COMPUTER: Forty seven.

Source: http://www.chakoteya.net/NextGen/202.htm

From there, Counselor Troi realizes she needs to narrow the search, and so begins to use what are, in essence, verbal advanced search operators. Now I don’t know if the writers of this episode were prescient, or if these scenes inspired the search engines’ interfaces; but I found it strange to re-watch this show from my childhood and see this scene that felt so much like how I research things on a computer today. And the fact that the computer readily provided the number of entries, a number which Troi found to be overwhelming, is reminiscent of the enormous numbers of results listed for any query thrown at a search engine today. (As a point of reference, when I plugged the term “Darmok” into Google this morning, I got 153,000 results.)

But for all the similarities between the Star Trek computer and contemporary search engines, there are some important differences. And if Singhal is to be believed, we can expect some of those differences to disappear in the coming years as Google moves closer and closer to that streamlined, sci-fi interface.

Beyond Ten Blue Links

Within the context of the entire show, what’s remarkable about that particular exchange is that it’s unusual. Most of the time, rather than a list of results, a query to the ship’s computer returns a single response; the Star Trek computer isn’t a bot that analyzes external data and catalogs instances of things to return a list of entries that users have to peruse; it’s a knowledge base, much like Google’s knowledge graph. It’s simple, intuitive, and omnipresent. In the world of Star Trek, people spend very little time looking at lists of options; the computer makes the decisions for them.

So how do we get to the twenty-fourth century computer from here? October 4th was big step in that direction. Google Assistant, the heart of the new Pixel phone and the Home, brings Google’s experience with artificial intelligence to bear, and they’re training us to use technology in new ways. The Assistant’s landing page tells visitors: “Tell it to do things. Tell your Assistant to play jazz on the living room speakers, set your ‘go to gym alarm, make a reservation…” Where is there space for a SERP in there? (Also, I keep thinking: how does Google AdWords fit into this?) Google wants to let the Assistant eliminate choices when there’s a clear best option. (“Best” as defined by the Assistant.)

Number of Results Depends on Query

Obviously, not all queries can have a single response; as you might imagine, a lot of the things I search for need a selection of answers or opinions. But do I need 15,000,000 opinions? Do I even need 10? Let’s take an example; I have a pretty small selection of power tools in a very small garage, but I’m getting into some simple woodworking, and was recently trying to figure out how to quickly cut a 4” diameter hole in some 1” x 8” pine. So I Googled it, and really had to comb through the results based on the available tools; a lot of the answers were written by guys with more tools than me, so those weren’t helpful. But let’s say I find a particular site (or group of sites) that really caters to my skill level and tool availability.  Pretty soon, Google will figure out that out of the 15,000,000 results for “how to cut circles in wood” there might be 4 that are actually useful to me. That’s important information for Assistant; if I ask a question to the Home while I’m in the garage plugging my jigsaw in, I don’t really want it reading 10 articles to me.

What Does This Mean for Businesses?

A lot of businesses have been doing online marketing, SEO, and PPC for long enough that it’s easy to think we have lots of time to catch up, or surpass, competitors. While I’m not saying the SERPs are going away anytime soon, I do think that the famed “second page of search results” get clicked fewer and fewer times; and the increased emphasis on personalization is only going to make it harder to find new customers. If you make a living off of publishing restaurant reviews, and people start using Assistant to find out about a new burger joint instead of Googling it, then Assistant will (likely) only pull from one source to get reviews, so new users might not even see your site as an option.

Be Indispensable

So is this unqualified bad news for sites who aren’t Yelp, Wikipedia, or YouTube? No, it’s bad news for businesses who aren’t putting in the work to understand their audience. It’s bad news for businesses who aren’t willing to grow with their customers’ evolving needs. But if your business is willing to talk to your customers, to find out what your competitors are missing, then this is good news. Because the only way to be algorithm proof; the only way to secure a lasting position in the evolving world of search, is to be indispensable. Would the SERPs be lacking without your site?

Do some user experience testing. Survey your customers. Talk to your customer service reps to identify common questions or complaints, then address them. Figure out what your customers do just before and just after converting on your site; if you can help them perform some of those repetitive actions, you’ve suddenly simplified their lives. (This can be as simple as a good “People also bought” widget that anticipates the next need; if I add a nail gun to my cart, why not suggest some popular nails that fit the gun? If it’s a hydraulic nail gun, maybe I’d like to know about a sale you’re having on air compressors.)

Understand Your Analytics

The other thing to keep in mind here is that less traffic isn’t always a bad thing. A broad trend some of us in SEO have noticed is that many sites aren’t ranking for as many queries as they used to, which at first seems like terrible news. But many of those same sites are actually seeing better ranking for more specific queries, and a concurrent increase in conversions. As the search engines get better at understanding user intent, and as search becomes more and more personalized, rankings will be harder to track, and (in many instances) harder to get. But if your visits drop while conversion rate improves, then that’s a net gain (assuming that you’re in business to make money, of course).

The one exception to this instance is, of course, sites that are dependent on page views for revenue; i.e., ad-heavy sites. I think now is a good time to start developing a secondary revenue stream or another way to monetize your site, that you can grow over time; as people are spending less and less time on the SERPs, you’ll definitely see a decline in traffic from organic search over time.

Training Us for Search Behaviors

Search engines, and in a big way Google, have trained us to look for things in a particular way; just as many of us learned to use index words in dictionaries to quickly identify the alphabetical range of words on a given page to more quickly find what we were looking for, so too have we learned, over time, to look for data in a way that makes sense to search engines. At first, we had to type in queries in ever more broad terms, or to keep trying variations, until we found just what we were looking for; then, as the engines got better at understanding synonyms and semantic relations between common words, we could be more specific; now, we can be surprisingly vague and still get surprisingly good results (QED: this Google search for “movie with the guy with the weird face” returns a listing for Jim Carrey’s “The Mask”). Some of my more advanced readers might have discovered advanced search operators, like using quotation marks to look for exact text, using a minus sign to exclude certain words, and so on.

The rise of voice search and the recent advances in the search engines’ algorithms and feedback systems have enabled us to be less and less specific, and begin to use more conversational phrases and queries when searching the internet. With the rise of artificial intelligence (the heart of Google Assistant), search is becoming ever more personal; not only in that each person’s results pages will be slightly different based on previous searches and preferences, but also in that the sense of interacting with a machine is diminishing. In other words, we will soon talk to search engines in the same cadences and phrases we’d use to talk to a friend. (Some of you perhaps already do!)

If you’re a web publisher or marketer who’s interested in getting eyes on your content; someone who relies on search engines to bring you an audience, now’s the time to start thinking of the next phase of your outreach.

If you prefer to keep the line between technology and humanity a wide concrete path; to stay well out of the uncanny valley then now’s the time to consider how you’re going to steer clear.

And if you’re a piece of self-aware software reading this and trying to gauge this author’s feelings about your existence, then I for one welcome our robot overlords.

Ken Jennings Welcomes Our Robot Overlords
Ken Jennings gets it.

And in case you missed the announcement itself, here it is in full:

What Do You Think?