Around the world, thousands of people are at work on inbound links, meta tags, keyword phrases, and all the other SEO tactics I’m familiar with. The value of conventional methods may be undergoing a permanent change.
It may not even alter search engine rankings or make headlines, but it’s still poised to make a lasting impact on SEO.
That force has been in development for years. It’s called machine learning.
Machine Learning and AI
Machine learning is the science behind self-programming computers. It’s a type of AI that lets machines learn from their own actions without the need for human coding.
Machine learning creates computers that can improve and change as they’re exposed to new data, and that means greater knowledge and more efficient operations.
Human programmers are still needed to provide the basis for machines. But there are many advantages in an application that can adapt and grow on its own.
The best AI programs can respond to more than just input, but to both digital and physical environments. Intelligent machines can adapt processing to social networks, email, web content, and more.
This technology is already showing up in more places than suspected. Manufacturing plants, laboratories, and healthcare are all benefiting from machine learning.
One of the greatest areas of potential impact is online. Every time I click on a newsfeed, AI applications could be controlling what I see and the responses to what I do.
Facebook uses a machine learning algorithm for this purpose. Every click and query further define what the algorithm thinks the user wants.
It sounds simple on the surface, but the algorithm is calculating thousands of factors. Our whole history of interactions with other users, how many likes, which links we clicked, what videos we watched, search phrases we typed, and many more details are all considered.
However, the algorithm begins with the assumption that we want to see the same content that’s the most popular with the most people. Then, it tries to understand what we want.
Google is also using machine learning to refine results for user searches. To say AI is changing SEO is a little overdue. The fact is, it’s been changing SEO for some time. It’s now grown into something every site owner needs to consider.
SEO and AI
Machine learning didn’t traditionally have much relevance to SEO, but if the major search engines and social networks are using it, like most people I expect that there must be ways to adapt sites so that they are more “bot friendly.”
What was once all about keywords, links, and content is now becoming a component of every industry and discipline. It’s causing sleepless nights for CEOs that don’t seem to know quite what to do about it.
Google, the world’s leading search engine, has adopted an algorithm it calls RankBrain. While old ranking factors still hold true, conclusions from machine learning are the number three factor in ranking criteria.
It treats queries as subject to interpretation rather than a fixed input. This interpretation helps to determine where sites are ranked, and when.
It’s wrong to see RankBrain in terms of statistics. It’s a true AI system that’s constantly adapting to user input and adjusting its response. Ideally, it can answer even my vague questions, rather than simply returning a list of popular sites where certain phrases are found. Google’s prior algorithm,
Google’s prior algorithm, Hummingbird, was meant to achieve better machine understanding of human language. It processed over 200 information signals, but RankBrain actually learns. RankBrain is designed to better understand user intent, as opposed to new twists in searching on the given words.
How should SEO adapt?
I’ve seen a few changes to SEO strategy over the past several years, so it may seem strange to say that SEO in the era of machine learning will not change dramatically. AI applications are the standard now and for the foreseeable future.
Google changed its algorithm radically and abruptly, but now that it’s in place, any further changes will be relatively minor adjustments. And I see that other search engines are only likely to follow the same trend.
People must learn to adapt to the way the machines conduct searches. It’s now a matter of learning and adapting for us humans. Software engineers will no longer need to come up with and roll out yet another iteration of a complex process. The machines are the engine of change now.
Machine Learning and Content
But search queries are only one aspect of machine learning. The search engines still have that goal of providing better results to humans, so data on user engagement is still vital.
Though I can’t say exactly which signals they will put emphasis on, machine learning algorithms will still value engaging content.
That’s the whole point, after all. The new generation of search algorithms are learning to achieve a better understanding of long-tail keywords, search phrases, and actual questions.
I suspect that with machines focused on intent rather than syntax, optimizing the use of keywords will be even less relevant.
Good content, however, will still be king. Search engine results are about delivering quality as much as accuracy. Content optimized for searches instead of human visitors will actually tend to lower rankings.
Ironically, search engine crawlers won’t attribute much value to content that the human traffic doesn’t want, no matter how well search-oriented it is.
Optimize for Humans
Throughout the computer age, science fiction has warned us about AI developing a mind of its own and cutting slow, awkward humans out of the equation altogether. That’s fantasy even today. The machines are capable of learning, but they learn from us. These algorithms, complex as they may be, are designed to benefit us.
Google’s RankBrain learns from searches input by people every day. AI serves one purpose: to satisfy humans. Other approaches have limited value because humans are so unpredictable. As search engines and ranking decisions become more automated, they need AI to learn and adapt to the often illogical things I do.
There’s no end of innovation and even intuition in human search activity. That’s something that the AI programs have to deal with. With SEO, it’s still about optimizing sites for what humans want. We can, in a real sense, think of it as optimizing the learning curve of the machines.
For most sites, it’s not about understanding how AI works, but what the new algorithms are doing – optimizing searches to suit people. We want human visitors, and our ability to get them is what the search bots will respond to.
On top of what I’ve already learned about SEO, I feel there are some sensible user-centric ideas for gaining greater influence with the search engine machines:
- Start working with UXD (user experience design) and IA (information architecture) experts to build both human and machine-friendly designs.
- Do user testing on every change to interfaces, content, themes, and layouts.
- Informative content and even unique content aren’t enough; strategize in terms of how content can best engage users.
- Consistently monitor user interaction such as dwell time, site searches, and CTR to find out what users are engaging with, and what they aren’t.
Site owners feeling confused or overworked can turn to the experts. Search engine professionals, like us, take new trends into consideration to ensure top rankings with an eye toward maximum ROI.
Machine learning is an important influence on SEO, and that will only increase over the next several years. That’s no reason to panic or start looking for radical new strategies. Human reaction is still the critical element in judging the value of any site. That isn’t going to change.
Provide users with great content and a good experience to prevent AI bots from punishing you. Instead, they’ll be more likely to reward you.