Artificial intelligence (AI) is penetrating every department of every industry, from automating factory work to improving areas previously thought untouchable by machines (like human resources). But as a veteran in the online marketing world, I can’t help but let my imagination wander on how AI and machine learning are going to impact the world of search engine optimization (SEO)—the strategies organizations use to rank higher in search engine results pages (SERPs).
Already, we’re seeing the beginnings of a full-scale AI revolution in SEO, and search marketers are scrambling to keep pace with the changes. But what will the next few years bring? What about the next decade?
The Big Picture
We say “search engines,” but most of the time, we’re talking about Google. Bing, Yahoo!, DuckDuckGo, and other engines only share a fraction of the search user base, and most of their systems are modeled after Google’s in the first place. So our big question is, how is Google going to incorporate AI in the future to change how search works for the average user?
Historically, Google has updated its algorithms with two primary goals in mind:
- Improve user experience. Google wants users to find the answers they’re looking for, and receive accurate, valuable content. This is an important category, and a complicated one; to achieve this, Google not only has to perfect how its search engine functions, but also how it finds, organizes, and evaluates the quality of content on the web.
- Keep users on Google. Google makes money when people use it, and stay on the platform as long as possible. We’ll see why that’s important in a future section.
Google is already making use of machine learning in a few different ways, and it’s only a matter of time before it advances.
RankBrain and Machine Learning
First, let’s consider RankBrain, a machine learning-based upgrade to Google’s Hummingbird algorithm, which launched in 2015. The Hummingbird update, from 2013, originally rolled out “semantic search” capabilities. It was designed to evaluate the context of user queries, rather than the exact contents; rather than prioritizing exact match keywords, Hummingbird allowed Google to consider synonyms, related phrases, and more. This was a step in the right direction, because it meant users could find better results, and search optimizers could no longer get away with keyword stuffing.
RankBrain was a modification that allowed Google to study massive quantities of user search data and automatically improve its interpretation of user phrases. It was primarily focused on long, convoluted, or hard-to-understand phrases, ultimately reducing them down to a length and simplicity level the algorithm could more easily handle. It’s been self-updating and improving ever since.
This is an important indication of how search will evolve in the future; I’m guessing that rather than seeing manual update after manual update, we’ll see more algorithm changes designed to self-update based on machine learning insights. This is much faster and less costly than having humans doing all the work.
Content Quality and Link Quality
I suspect we’ll also see major AI advancements applied to better understand the quality of the content and links produced by search optimizers.
Links and content are the focal points of most SEO strategies. Google studies links to calculate domain- and page-level authority (or trustworthiness); generally, the more links a site has pointed to it, and the better those links are, the higher it’s going to rank. Similarly, better-written, more relevant content tends to rise in SERP rankings—and appeal to web users. Better content and better links mean you’ll end up with a higher return on investment (ROI) for your SEO strategy.
Over the years, Google has gotten better at analyzing the quality of content and links from websites; search marketers have evolved from trying to trick Google’s algorithm to simply trying to produce their best possible work.
Right now, Google’s methods for evaluating the subjective “quality” of content and links are good—but they could always be better. It would be easier for an AI agent to gradually learn what makes good content “good,” than to rely on a manual agent coding those parameters into a system. I believe Google will make more efforts to automate quality evaluation in the near future.
Individualization
Google has also taken great efforts to individualize its search results. If you search for the same phrase in Phoenix, Arizona and Cleveland, Ohio, you’re probably going to get radically different results. You may also get different results based on your search history, and even the demographic information Google “knows” about you.
Right now, these individualization efforts are impressive, but limited. We’re not surprised that Google knows where we are, or the last few things we’ve searched for. But in the near future, Google may be capable of using AI to make more intensive predictions. Based on your historical searches and search data from millions of other users like you, Google may be able to recommend searches or search results before you even know you need them.
For search marketers, this is both an opportunity and a threat. If you can capitalize on predictive searches, you can get a massive edge on the competition—but then again, if Google’s algorithmic methods are opaque, you may have a hard time understanding how and when your results appear for users.
Smart Results
Over the past few years, Google has stepped up its efforts to keep users on the SERPs, rather than clicking links to visit other websites. The Knowledge Graph and rich snippets now appear to provide immediate answers to user queries, preventing the need to click any further. As Google gets better at dissecting user queries with RankBrain and Hummingbird, and becomes better at parsing the web with smart algorithms, I suspect we’ll see even more of these user-attention-grabbing entries.
For search marketers, this is again both an opportunity and a threat. If you can game the system and get your content to appear in the SERPs above your competitors’, you’ll get a major boost to your brand reputation. But at the same time, if users stay in the SERPs, and never visit, you’ll miss out on a ton of organic traffic.
Real Time Changes and Adaptability
AI is remarkably good at analyzing vast amounts of data, and far faster than even an experienced human team. Historically, Google has made periodic updates to its algorithm with major, game-changing algorithm changes dropped every few months. But recently, those algorithm updates have tapered off in favor of much smaller, much more frequent updates.
This trend will likely develop further in the future as Google’s AI systems optimize toward real-time analytics. It will “learn” constantly, with every new search query, and possibly roll new updates to its live algorithm on a constant basis, making it difficult to keep up with its iterative evolution.
Content Production and Onsite Optimization
It’s also worth noting that AI won’t just be harnessed by Google and other search engines. We’ll also see the development and utility of AI on behalf of search marketers. AI-based content generators are becoming more advanced and more common; eventually, search marketers may be able to use them to produce and distribute content good enough to “fool” Google’s algorithms. From there, this will likely turn into an arms race between search marketers and search algorithms—not too unlike what we already have.
Furthermore, smart onsite optimization engines could greatly simplify the technical efforts that search marketers currently have to make. Current plugins and onsite SEO tools are helpful, but incomplete; in the near future, AI and machine learning could make these substantially more capable.
Overall, it’s unlikely that we’ll see such a radical transformation that SERPs become unrecognizable, or that SEO disappears as an online marketing strategy. However, search marketers and users will both have to make some serious adjustments if they’re going to stay relevant as AI infiltrates this space.
