Bold predictions make good copy. I’m as guilty as the next person; asked for predictions about local search in 2017, for Street Fight and elsewhere, I found myself channeling the zeitgeist and suggesting that with the advent of voice search, we’d see an inevitable trend toward specific queries with specific answers, and thus the death of the ten blue links of search and all they have represented.
In this I was echoing David Mihm: “SEO will not die in 2017. But the number of beneficiaries of organic visibility from Google will decline dramatically. Many local marketing practitioners continue to use the adage ‘if you aren’t on the first page of Google results, you’re invisible.’”
That’s already out-of-date and about to be wildly out-of-date. We’ll soon be saying—maybe not by the end of this year, but soon enough—”if you’re not the first result on Google, you’re invisible.”
This notion of a trend toward one best answer to each question is born out of the growth of the Knowledge Graph and featured snippets, which demonstrate Google’s attempt to highlight the one best answer to certain questions, such as “how to jumpstart a car.”
An equal contributor to this sense of a trend is voice search and AIs like Alexa, Google Assistant, Cortana, and Siri, who promise a future where every need is understood and met perfectly with the right information or the appropriate service.
To an extent though, this dream of the perfect answer to every question has been muddled up with what we imagine has changed now that voice search has made the interface largely invisible. There seems to be a superficial assumption that voice assistants must always make their best attempt at a single answer to every query, simply because an interface with no visual component is incapable of showing you a list of options.
If this superficial assumption were true, then the job of local search marketing would be on the verge of a massive shift in complexity, where all the old rules would no longer apply. As Mihm writes, “If you’re not the first result on Google, you’re invisible.”
But how true is this likely to be? Consider the classic local search query: “Find me a restaurant.” Who wants one answer to that question? Wouldn’t we rather be presented with a range of choices so we can pick a place that meets our mood or the consensus of the group we’re in that evening?
How about queries for local service providers? In our house, the gas fireplace has begun acting up, so I need to find a gas fireplace repair specialist. I’ve never needed such a service provider before, so of course I’ll turn to local search to find the right candidate. But if I’m only provided with one option, I have no recourse if that result doesn’t meet my needs, if they’re booked for a month or too expensive or they don’t service my area. I will want a range of options and I might easily end up hiring the second or third specialist on the list.
Indeed I tried several local search queries on Alexa via Amazon Echo and Google Assistant via Allo, just to test this theory, and for the most part it’s true that the AI presents a range of options for the user to choose from. I tried simple queries like “Find me a restaurant” and more complex queries like “Find me a dentist who specializes in emergency crown repairs.” Interestingly, the complex queries performed very poorly in my tests, often baffling the AI, whereas the simpler queries tended to produce results that look very familiar to those who are used to Google and Apple Maps or Google search with the three-listing Snack Pack.
Below is a table that tallies the results and their format where applicable. Note that all of Alexa’s results are powered by Yelp and shown visually in the Alexa app in response to the user’s voice query.
As you can see, the search performance for both Alexa and Google was highly variable, but some patterns are clear. Alexa is having growing pains as a local search service, not unlike those suffered by Apple in the early days of Apple Maps. Google shows itself to be much more savvy when it comes to providing relevant local results, though both services are not as clever as you’d expect in understanding the intent of complex long-tail queries.
But for simple queries that were well understood by the AI, both services responded in a manner that is surprisingly reminiscent of organic and local search on the desktop and in mobile devices. That is, they attempted to provide relevant results ranked by factors like relevance, popularity, and distance, and they almost never returned a single result for any local query.
See for example the familiar Snack Pack result for my gas fireplace repair query.
My informal study would appear to demonstrate that at least for now, Alexa and Google are thinking of AI-powered local search in the traditional sense of providing the user with a range of relevant options, even when organic search is trending toward the single best answer.
This isn’t to say that ranking will not become more complex. These new interfaces will use a different approach to natural language parsing that does not necessarily follow the “keyword plus location” model we have been trained to think of as the paradigm for local. Ranking factors may become hopelessly impossible to measure. But I think it’s more likely that at least many of the same indicators of relevance – keyword matching, popularity, proximity, and the rest – will continue to determine whether a business is surfaced in search.